CN107220962A - A kind of image detecting method and device of tunnel crackle - Google Patents

Abstract

The invention discloses an image detection method and device for tunnel cracks. The method includes: performing bilateral filtering on the image to be detected of the tunnel crack to obtain the filtered image; using a visual saliency model to respectively construct a luminance saliency map of the filtered image and a texture saliency map of the filtered image; fusing the luminance saliency map and the texture The saliency map is obtained to obtain a fused saliency map; the fused saliency map is segmented by an adaptive threshold algorithm to obtain a crack area image; when the crack in the crack area image is determined to be a real crack, the crack parameters of the crack area image are obtained. According to the image detection method provided by the embodiment of the present invention, cracks with poor continuity and low contrast can be effectively detected.

Description

Image detection method and device for tunnel cracks

technical field

The invention relates to the technical field of image processing and identification, in particular to an image detection method and device for tunnel cracks.

Background technique

With the wide application of image acquisition equipment such as digital cameras, video cameras, and ultra-high-speed scanners, and the continuous improvement of industrial production technology and production efficiency, there are also higher and higher requirements for the supporting production and testing capabilities. With the increasing development of image processing technology, image detection technology is widely used in industrial production process detection, daily life safety detection and other fields, which greatly improves the production efficiency of enterprises and people's living standards.

In the technical field of image processing and recognition, in general, the detection of industrial safety usually uses an image segmentation method to detect and recognize a specific area in the image. The existing image segmentation methods mainly use the overall gray level difference between the region of interest and the background region, and select an appropriate threshold to segment the image to obtain the region of interest. When the illumination is uneven or the gray level difference between the area to be detected and the background area is small, it is often impossible to accurately segment the area of interest.

Contents of the invention

Embodiments of the present invention provide an image detection method and device for tunnel cracks, which can effectively detect cracks with poor continuity and low contrast.

According to an aspect of the embodiments of the present invention, an image detection method for tunnel cracks is provided, the image detection method includes: performing bilateral filtering on the image of the tunnel crack to be detected to obtain the filtered image; using the visual saliency model to construct the filter The luminance saliency map of the filtered image and the texture saliency map of the filtered image; the fused saliency map and the texture saliency map are fused to obtain a fused saliency map; the fused saliency map is obtained by segmenting the fused saliency map through an adaptive threshold algorithm to obtain a crack area image; When the crack is a real crack, the crack parameters of the crack area image are obtained.

According to another aspect of the embodiments of the present invention, an image detection device for tunnel cracks is provided, the image detection device includes: an image filtering module, which is used to perform bilateral filtering on the image of the tunnel crack to be detected to obtain a filtered image; The graph construction module is used to construct the luminance saliency map of the filtered image and the texture saliency map of the filtered image respectively by using the visual saliency model; the saliency map fusion module is used to fuse the luminance saliency map and the texture saliency map to obtain the fused saliency map; The saliency map segmentation module is used to segment and fuse the saliency map through the adaptive threshold algorithm to obtain the crack area image; the crack parameter acquisition module is used to obtain the crack parameters of the crack area image when the crack in the crack area image is determined to be a real crack.

In the image detection method and device in the embodiments of the present invention, by constructing the luminance saliency map and the texture saliency map of the image to be detected, and fusing the luminance saliency map and the texture saliency map to obtain the fused saliency map, the images with poor continuity and low contrast can be Cracks are highlighted and enhanced in the image, and the obtained fused saliency map is segmented, and the crack parameter information is determined and counted, so that tunnel cracks can be effectively detected.

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings that need to be used in the embodiments of the present invention. Additional figures can be derived from these figures.

FIG. 1 is a flowchart illustrating an image detection method for tunnel cracks according to an embodiment of the present invention;

FIG. 2 is a schematic structural diagram showing an image detection device for tunnel cracks according to an embodiment of the present invention;

Fig. 3 is a structural diagram showing an exemplary hardware architecture of a computing device capable of implementing the image detection method and apparatus for tunnel cracks according to an embodiment of the invention.

detailed description

The characteristics and exemplary embodiments of various aspects of the present invention will be described in detail below. In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only configured to explain the present invention, not to limit the present invention. It will be apparent to one skilled in the art that the present invention may be practiced without some of these specific details. The following description of the embodiments is only to provide a better understanding of the present invention by showing examples of the present invention.

It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is a relationship between these entities or operations. any such actual relationship or order exists between them. Furthermore, the term "comprises", "comprises" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article, or apparatus comprising a set of elements includes not only those elements, but also includes elements not expressly listed. other elements of or also include elements inherent in such a process, method, article, or device. Without further limitations, an element defined by the statement "comprising..." does not exclude the presence of additional same elements in the process, method, article or device comprising said element.

The method and device for image detection of tunnel cracks according to the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that the embodiments described in the present invention are not intended to limit the scope of the present disclosure.

FIG. 1 is a flow chart illustrating an image detection method for tunnel cracks according to an embodiment of the present invention. As shown in FIG. 1 , the image detection method 100 for tunnel cracks in this embodiment includes the following steps:

Step S110, performing bilateral filtering on the image of the tunnel crack to be detected to obtain a filtered image.

Step S120, using the visual saliency model to respectively construct a brightness saliency map of the filtered image and a texture saliency map of the filtered image.

Step S130, fusing the luminance saliency map and the texture saliency map to obtain a fused saliency map.

Step S140, segmenting and merging the saliency map through an adaptive threshold algorithm to obtain a crack area image.

Step S150, when it is determined that the crack in the crack region image is a real crack, obtain the crack parameters of the crack region image.

According to the image detection method of the embodiment of the present invention, tunnel cracks can be effectively detected and better detection results can be obtained.

As an optional embodiment, the steps of bilateral filtering processing in step S110 may specifically include:

Step S111, counting the gray variance value of each pixel in the specified neighborhood in the image to be detected.

In this step, the specified neighborhood can be, for example, a rectangular window with a size of M×N pixels, and the value of M or N can be an odd number between 7 and 21.

Step S112 , calculating the average grayscale variance value of the image to be detected through the grayscale variance value of each pixel.

Step S113 , setting a grayscale variance parameter according to the average grayscale variance value, and constructing a bilateral filter kernel function based on the grayscale variance parameter.

Step S114, using the constructed bilateral filter kernel function to perform bilateral filtering on the image to be detected through the convolution template formula, and use the image to be detected after bilateral filtering as a filtered image.

For ease of understanding, as an example, the embodiment of the present invention calculates the average gray variance σ of the image to be detected, and uses σ _r =σ/2 as the gray variance parameter, and uses the following formula (1) to construct a bilateral filter kernel function.

In the above formula (1), ω(i, j, k, l) represents the weight coefficient of each pixel in the specified neighborhood, parameter i and parameter j represent the coordinates of the image pixel in the specified neighborhood, parameter k and parameter 1 represents the coordinates corresponding to the parameter k and the Gaussian template to which parameter 1 belongs, σ _d represents the variance of the spatial Gaussian template, and σ _r represents the variance of the Gaussian template in the range. In bilateral filtering, the weight coefficient ω(i, j, k, l) is the product of the range Gaussian template ω _r (i, j, k, l) and the spatial domain Gaussian template ω _d (i, j, k, l) , ie ω(i, j, k, l)=ω _d (i, j, k, l)×ω _r (i, j, k, l), and the range Gaussian template Spatial Gaussian Mask

Using the constructed bilateral filter kernel function (1), use the following formula (2) to perform bilateral filter filtering:

In the above formula (2), I'(i, j) represents the image data obtained after filtering by the bilateral filter.

In this embodiment, the bilateral filtering method is used to preprocess the image to be detected, which can eliminate the noise of the image to be detected while maintaining image edge information, smooth the image to be processed, and have an obvious protection effect on image details.

In the embodiment of the present invention, the visual saliency model is an algorithm proposed based on the human visual attention mechanism. The human visual attention mechanism means that human visual attention usually focuses on abruptly changing brightness regions or texture regions in visual signals.

In some embodiments, the step of constructing the brightness saliency map of the filtered image and the texture saliency map of the filtered image in step S120 may further include:

Step S121 , according to the visual saliency model, construct a brightness Gaussian pyramid of the filtered image, and the brightness Gaussian pyramid includes a predetermined number of layers.

In the embodiment of the present invention, as a multi-scale image representation method, the Gaussian pyramid can efficiently extract the features of the image at different scales. The process of constructing the brightness Gaussian pyramid in the present invention will be described in detail below.

First, extract the luminance feature in the original image to be processed to obtain the original image of luminance feature, and use the original image as the bottom image of the pyramid; then, the image after Gaussian function filtering of the original image is used as the second layer of the luminance Gaussian pyramid image, and the length of the second-layer image is 1/2 of the length of the bottom image, and the width of the second-layer image is 1/2 of the width of the bottom image; The third-level image of the Gaussian pyramid, and the length of the third-level image is 1/2 of the length of the second-level image, the width of the third-level image is 1/2 of the width of the second-level image, ..., and so on , to construct a brightness Gaussian pyramid of the image after continuous Gaussian filtering. In addition to the bottom layer of the brightness Gaussian pyramid, the length of each layer image from high to low according to the number of layers is 1/2 of the length of the next layer image adjacent to this layer, and the width of each layer image is 1/2 of the length of the layer image. 1/2 of the image width of the adjacent next layer.

As an example, the number of layers of the brightness Gaussian pyramid constructed in the embodiment of the present invention is 5 layers.

Step S122, using the Gabor filter function to obtain the texture Gaussian of the filtered image by calculating the Gabor filter image of each layer image of the brightness Gaussian pyramid in the four directions of 0, π/4, π/2, and 3π/4 pyramid.

Specifically, the Gaber filter image of each layer of the brightness Gaussian pyramid in the four directions of 0, π/4, π/2, and 3π/4 is calculated by the following formula (3) to form a texture pyramid:

In the above formula (3), is the kernel function of the Gabor filter, and the kernel function of the Gabor filter has the following parameters: wavelength λ, direction θ, phase shift Standard deviation σ, aspect ratio γ, x and y are the coordinates of the image pixel of the current layer of the brightness Gaussian pyramid.

Step S123, according to the number of layers of the luminance Gaussian pyramid from high to low, obtain images of each layer except the bottom layer as the image to be processed, and perform up-sampling on the image to be processed to obtain the image of the next layer whose resolution is adjacent to the image to be processed images with the same resolution, and calculate the absolute value of the image subtraction operation between the image obtained by upsampling and the image of the adjacent next layer image to obtain a brightness saliency map.

Step S124, according to the number of layers of the texture Gaussian pyramid from high to low, obtain each layer image except the bottom layer as the image to be processed, and perform up-sampling on the image to be processed to obtain the next layer image whose resolution is adjacent to the image to be processed images with the same resolution, and calculate the absolute value of the subtraction operation between the image obtained by upsampling and the image of the adjacent next layer image to obtain the texture saliency map.

In above-mentioned step S123 or step S124, utilize following formula (4) to carry out image subtraction operation, and obtain the absolute value to the result of image subtraction operation:

s(i,j)=|p(i)-p ^T (j)| (4)

In the above formula (4), p(i) represents the pyramid image of the i-th layer, and p ^↑ (j) represents an image with the same resolution as the i-th layer image obtained by upsampling the j-th layer pyramid image , and i<j.

In the embodiment of the present invention, image subtraction refers to a point-to-point subtraction of pixel values between pixels of two or more images. Taking image subtraction of two images as an example, each pixel of one of the images is obtained as the first pixel, and the pixel value of the first pixel is compared with the pixel value of the other image and the first pixel. The pixel values of the pixels at the same pixel position are subtracted to obtain an image subtraction result of one of the images and the other image.

Through the image subtraction operation in the embodiment of the present invention, the unnecessary superimposed pattern in the image to be detected can be removed when the crack is detected, and when the crack contrast is low and the continuity is poor, the background image can be effectively removed to make the crack display effect be strengthened, so as to obtain a better detection effect in the subsequent crack detection.

In some embodiments, the step of fusing the brightness saliency map and the texture saliency map in step S130 may specifically include:

Step S131, obtain the maximum pixel value of the pixel point in the luminance saliency map as the first pixel value, obtain the maximum pixel value of the pixel point in the texture saliency map as the second pixel value, and combine the pixel value of the pixel point in the luminance saliency map with The first pixel value is divided to obtain a normalized brightness saliency map, and the pixel value of a pixel point in the texture saliency map is divided by the second pixel value to obtain a normalized texture saliency map.

Specifically, the normalization process of the brightness saliency map or the texture saliency map is performed by the following formula (5):

s _n (i, j) = s (i, j)/M (i, j) (5)

When using the above formula (5) to normalize the luminance saliency map, s _n (i, j) represents the normalized luminance saliency map obtained after normalizing the luminance saliency map, s(i, j ) is the pixel value of a pixel in the luminance saliency map, and M(i, j) is the maximum pixel value of a pixel in the luminance saliency map.

When using the above formula (5) to normalize the texture saliency map, s _n (i, j) represents the normalized texture saliency map obtained after normalizing the texture saliency map, s(i, j ) is the pixel value of the pixel in the texture saliency map, and M(i, j) is the maximum pixel value of the pixel in the texture saliency map.

Step S132, comparing the pixel value of each pixel in the normalized brightness saliency map with the pixel in the same position as each pixel in the normalized texture saliency map, and using the maximum value obtained from the comparison as the fusion The pixel value of the pixel in the same position as each pixel in the saliency map is obtained to obtain the fused saliency map.

Specifically, the brightness saliency map and the texture saliency map are fused by the following formula (6):

In the above step (6), is the normalized brightness saliency map obtained after normalization processing, is the normalized texture saliency map obtained after normalization processing.

In some embodiments, the adaptive threshold algorithm in step S140 is the maximum inter-class variance method, and step S140 may also include:

In step S141, the optimum crack segmentation threshold for the fused saliency map is calculated by the maximum between-class variance method.

In the embodiment of the present invention, the maximum inter-class variance method is also called OTSU algorithm or Otsu algorithm, and the maximum inter-class variance method is a method for maximizing the inter-class variance between the segmented foreground image area and the background image area.

Specifically, the selection of the optimal crack segmentation threshold can be expressed by the following formula (7):

T＝arg max _t (var(I<t)+var(I≥t)) (7)

In the above formula (7), var(I<t) represents the variance of the image area whose gray value is smaller than the t value relative to the overall gray value mean, and var(I≥t) represents the gray value when the gray value is not less than the t value The variance of the image area relative to the mean value of the overall gray scale, t represents the segmentation threshold, and T represents the value of t when var(I<t)+var(I≥t) takes the maximum value.

That is to say, a crack segmentation threshold is used to divide the entire image data into two classes. If the variance between the two classes is the largest, then this crack segmentation threshold is the optimal crack segmentation threshold.

In step S142, the optimal crack segmentation threshold is used to segment the fused saliency map to obtain a candidate crack region image of the fused saliency map.

In this embodiment, the optimal crack segmentation threshold is obtained by using the method of maximum between-class variance, which can minimize the probability of misclassification of the image to be detected by applying the optimal crack segmentation threshold, thereby improving the number of candidate cracks obtained after image segmentation. The accuracy of the area image.

In some embodiments, step S150 may specifically include:

Step S151 , respectively calculating the mean value of the gray value of the pixels in the image of the crack area and the mean value of the gray value of the pixels in a specified area adjacent to the image of the crack area.

Step S152, if the average gray value of the pixels in the crack area is smaller than the average gray value of the pixels in the specified area, it is determined that the crack in the crack area is a real crack.

In this step, the average gray value of the pixels in the adjacent area of the candidate crack area obtained by statistical segmentation is used for the low gray value of the crack. If the average gray value of the crack is smaller than the average gray value of its adjacent image, the Cracks are real cracks.

Step S153, binary segmenting the crack region image to obtain a crack binary image of the crack region image, and performing a skeletonization operation on the crack binary image to obtain crack parameters of the crack region image, where the crack parameters include crack length and crack width.

Image skeletonization is a method for line image analysis. By extracting the skeleton of cracks judged as real cracks, the size and shape information of cracks, such as the length and width of cracks, are identified and counted.

According to the image detection method provided by the embodiment of the present invention, the crack region in the image can be detected and identified accurately, and good detection effect can also be obtained when the crack continuity is poor and the contrast is low.

The image detection device for tunnel cracks according to the embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings.

Fig. 2 shows a schematic structural diagram of an image detection device for tunnel cracks according to an embodiment of the present invention. As shown in FIG. 2, the image detection device 200 in the embodiment of the present invention includes:

The image filtering module 310 is configured to perform bilateral filtering on the image of the tunnel crack to be detected to obtain a filtered image.

The saliency map construction module 320 is configured to respectively construct a brightness saliency map of the filtered image and a texture saliency map of the filtered image by using the visual saliency model.

The saliency map fusion module 330 is used to fuse the brightness saliency map and the texture saliency map to obtain the fused saliency map.

The saliency map segmentation module 340 is configured to segment and fuse the saliency map through an adaptive threshold algorithm to obtain a crack region image.

The crack parameter acquisition module 350 is configured to acquire crack parameters of the crack area image when it is determined that the crack in the crack area image is a real crack.

According to the image detection device provided by the embodiment of the present invention, the luminance saliency map and the texture saliency map of the image to be detected are fused to obtain a fused saliency map, and the obtained fused saliency map is segmented to determine and count the crack parameter information, so that the tunnel crack can be obtained effective detection.

In some embodiments, the image filtering module 310 may also include:

A grayscale variance value statistical unit 311, used to count the grayscale variance value of each pixel in the specified neighborhood in the image to be detected;

The average grayscale variance value calculation unit 322 is used to calculate the average grayscale variance value of the image to be detected through the grayscale variance value of each pixel;

The bilateral filter kernel function construction unit 323 is used to set the grayscale variance parameter according to the average grayscale variance value, and construct the bilateral filter kernel function based on the grayscale variance parameter;

The bilateral filter function calculation unit 324 is configured to use the constructed bilateral filter kernel function to perform bilateral filtering on the image to be detected through the convolution template formula, and use the bilaterally filtered image to be detected as a filtered image.

In this embodiment, the bilateral filtering can eliminate the noise of the image to be detected while maintaining the edge information of the image to make the image to be processed smooth.

In some embodiments, the saliency map construction module 320 may also include:

A brightness Gaussian pyramid construction unit 321, configured to construct a brightness Gaussian pyramid of the filtered image according to the visual saliency model, where the brightness Gaussian pyramid includes a predetermined number of layers;

The texture Gaussian pyramid construction unit 322 is used to use the Gabor filter function to calculate the Gabor filter image of each layer image of the brightness Gaussian pyramid in the four directions of 0, π/4, π/2, and 3π/4 to obtain A textured Gaussian pyramid of the filtered image;

The luminance saliency map construction unit 323 is used to obtain the image of each layer except the bottom layer from high to low according to the number of layers of the luminance Gaussian pyramid as the image to be processed, and perform up-sampling on the image to be processed to obtain a resolution adjacent to the image to be processed The image with the same resolution as the image of the next layer, and calculate the absolute value of the image subtraction operation between the image obtained by upsampling and the image of the adjacent image of the next layer to obtain a brightness saliency map;

The texture saliency map construction unit 324 is used to obtain the image of each layer except the bottom layer from high to low according to the number of layers of the texture Gaussian pyramid as the image to be processed, and perform up-sampling on the image to be processed to obtain a resolution adjacent to the image to be processed The resolution of the next layer image is the same as that of the next layer image, and the absolute value of the subtraction operation between the upsampled image and the image of the adjacent next layer image is calculated to obtain the texture saliency map.

In this embodiment, the cracks in the image to be detected are enhanced through the construction of the saliency map. When the contrast of the crack is low and the continuity is poor, the background image can be effectively removed, which provides a good data basis for subsequent crack detection and identification.

In some embodiments, the saliency map fusion module 330 may also include:

The feature map normalization processing unit 331 is configured to acquire the maximum pixel value of the pixel in the luminance saliency map as the first pixel value, acquire the maximum pixel value of the pixel in the texture saliency map as the second pixel value, and set the luminance salience The pixel value of the pixel point in the figure is divided by the first pixel value to obtain a normalized brightness feature map, and the pixel value of the pixel point in the texture saliency map is divided by the second pixel value to obtain a normalized texture feature map;

The fused saliency map construction unit 332 is used to compare the pixel value of each pixel in the normalized brightness saliency map with the pixel at the same position as each pixel in the normalized texture saliency map, and compare The obtained maximum value is used as the pixel value of the pixel in the same position as each pixel in the fused saliency map to obtain the fused saliency map.

In some embodiments, the saliency map segmentation module 340 may also include:

The optimal crack segmentation threshold calculation unit 341 is used to calculate and obtain the optimal crack segmentation threshold of the fusion saliency map through the maximum inter-class variance method;

The saliency map segmentation acquisition module 340 uses the optimal crack segmentation threshold to segment the fused saliency map to obtain the image of the candidate crack area of the fused saliency map.

In this embodiment, the segmentation threshold selected for image segmentation is the optimal crack segmentation threshold obtained by the maximum between-class variance method, and using the optimal crack segmentation threshold to segment the fused saliency map can effectively reduce the misclassification of cracks Rate.

In some embodiments, the crack parameter acquisition module 350 may also include:

A gray value calculation unit 351, configured to respectively calculate the mean value of the gray value of the pixels in the image of the crack area, and the mean value of the gray value of the pixels in a specified area adjacent to the image of the crack area;

The crack authenticity determination unit 352 is configured to determine that the crack in the crack area is a real crack if the average value of the gray value of the pixels in the crack area is less than the average value of the gray value of the pixel points in the specified area;

The crack parameter acquisition unit 353 is used for binary segmentation of the crack region image to obtain the crack binary image of the crack region image, and performs skeletonization operation on the crack binary image to obtain the crack parameters of the crack region image, and the crack parameters include crack length and crack width .

In this embodiment, the crack region obtained after segmentation is judged by the characteristic of low gray level of the crack, and after it is judged as a real crack, parameter information such as crack length and crack width is obtained.

Other details of the image detection device for tunnel cracks according to the embodiment of the present invention are similar to the image detection method for tunnel cracks according to the embodiment of the present invention described above in conjunction with FIG. 1 , and will not be repeated here.

The image detection method and apparatus for tunnel cracks according to the embodiments of the present invention described in conjunction with FIG. 1 and FIG. 2 may be implemented by a computing device detachably or fixedly installed on an application server device. FIG. 3 is a structural diagram illustrating an exemplary hardware architecture of a computing device capable of implementing the image detection method and apparatus for tunnel cracks according to an embodiment of the present invention. As shown in FIG. 3 , the computing device 300 includes an input device 301 , an input interface 302 , a central processing unit 303 , a memory 304 , an output interface 305 , and an output device 306 . Wherein, the input interface 302, the central processing unit 303, the memory 304, and the output interface 305 are connected to each other through the bus 310, and the input device 301 and the output device 306 are respectively connected to the bus 310 through the input interface 302 and the output interface 305, and then connected to the computing device 300 other component connections. Specifically, the input device 301 receives image input information from the outside (for example, an imaging device or a digital camera), and transmits the input information to the central processing unit 303 through the input interface 302; The execution instruction processes the input information to generate output information, temporarily or permanently stores the output information in the memory 304, and then transmits the output information to the output device 306 through the output interface 305; the output device 306 outputs the output information to the computing device 300 external to the user.

That is to say, the computing device shown in FIG. 3 can also be implemented to include: a memory storing computer-executable instructions; and a processor, which can implement the operations described in conjunction with FIGS. 1 to 2 when executing the computer-executable instructions. Image detection method and device for tunnel cracks. Here, the processor can communicate with an image acquisition module such as an image management system or an image sensor installed on the device to be inspected, so as to execute computer-executable instructions based on relevant information from the image management system and/or image sensor, thereby realizing the combination of FIG. 1 to 2 describe the image detection method and device for tunnel cracks.

It is to be understood that the invention is not limited to the specific arrangements and processes described above and shown in the drawings. For conciseness, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the sequence of steps after understanding the spirit of the present invention.

The functional blocks shown in the structural block diagrams described above may be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, it may be, for example, an electronic circuit, an application specific integrated circuit (ASIC), suitable firmware, a plug-in, a function card, or the like. When implemented in software, the elements of the invention are the programs or code segments employed to perform the required tasks. Programs or code segments can be stored in machine-readable media, or transmitted over transmission media or communication links by data signals carried in carrier waves. "Machine-readable medium" may include any medium that can store or transmit information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, and the like. Code segments may be downloaded via a computer network such as the Internet, an Intranet, or the like.

It should also be noted that the exemplary embodiments mentioned in the present invention describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above steps, that is, the steps may be performed in the order mentioned in the embodiment, or may be different from the order in the embodiment, or several steps may be performed simultaneously.

The above is only a specific implementation of the present invention, and those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the above-described systems, modules and units can refer to the foregoing method embodiments The corresponding process in , will not be repeated here. It should be understood that the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present invention, and these modifications or replacements should cover all Within the protection scope of the present invention.

Claims (12)

1. an image detection method of tunnel crack, it is characterized in that, described image detection method comprises: Bilateral filtering is performed on the image to be detected of the tunnel crack to obtain the filtered image; Constructing a brightness saliency map of the filtered image and a texture saliency map of the filtered image respectively by using a visual saliency model; fusing the luminance saliency map and the texture saliency map to obtain a fused saliency map; Segmenting the fused saliency map by an adaptive threshold algorithm to obtain a crack region image; When it is determined that the crack in the crack region image is a real crack, the crack parameters of the crack region image are obtained. 2. The image detection method according to claim 1, characterized in that said filtering the image to be detected of the tunnel crack to obtain the filtered image comprises: Counting the gray variance value of each pixel in the specified neighborhood in the image to be detected; Obtaining the average grayscale variance value of the image to be detected by calculating the grayscale variance value of each pixel; Setting the grayscale variance parameter according to the average grayscale variance value, and constructing a bilateral filter kernel function based on the grayscale variance parameter; Using the constructed bilateral filter kernel function, the image to be detected is subjected to bilateral filtering through a convolution template formula, and the image to be detected after the bilateral filtering is used as the filtered image. 3. The image detection method according to claim 1, wherein said utilizing a visual saliency model to respectively construct a luminance saliency map and a texture saliency map of the filtered image comprises: According to the visual saliency model, constructing a brightness Gaussian pyramid of the filtered image, the brightness Gaussian pyramid including a predetermined number of layers; Using the Gabor filter function, the texture of the filtered image is obtained by calculating the Gabor filter image of each layer image of the brightness Gaussian pyramid in the four directions of 0, π/4, π/2, and 3π/4 Gaussian pyramid; According to the number of layers of the luminance Gaussian pyramid from high to low, each layer of image except the bottom layer is respectively obtained as an image to be processed, and the image to be processed is up-sampled to obtain a lower resolution adjacent to the image to be processed An image with the same resolution as one layer of images, and calculating the absolute value of the image subtraction operation between the image obtained by the upsampling and the image of the adjacent next layer image to obtain the brightness saliency map; According to the number of layers of the texture Gaussian pyramid from high to low, each layer image except the bottom layer is respectively obtained as an image to be processed, and the image to be processed is up-sampled to obtain a lower resolution adjacent to the image to be processed An image with the same resolution as the image in one layer, and calculate the absolute value of the subtraction operation between the image obtained by the upsampling and the image in the adjacent next layer, to obtain the texture saliency map. 4. The image detection method according to claim 1, wherein said merging said brightness saliency map and said texture saliency map to obtain a fusion saliency map comprises: Respectively obtain the maximum pixel value of the pixel point in the brightness saliency map as the first pixel value, obtain the maximum pixel value of the pixel point in the texture saliency map as the second pixel value, and set the pixel value of the pixel point in the brightness saliency map dividing the pixel value by the first pixel value to obtain a normalized brightness saliency map, and dividing the pixel value of a pixel point in the texture saliency map by the second pixel value to obtain a normalized texture saliency map; Comparing each pixel in the normalized brightness saliency map with the pixel value in the normalized texture saliency map at the same position as each pixel, and comparing the obtained maximum The value is used as the pixel value of the pixel at the same position as each pixel in the fused saliency map to obtain the fused saliency map. 5. image detection method according to claim 1, is characterized in that, described self-adaptive threshold value algorithm is the maximum variance method between classes; The step of segmenting the fused saliency map through an adaptive threshold algorithm to obtain an image of a candidate crack region includes: Calculate and obtain the optimal crack segmentation threshold of the fused saliency map through the maximum inter-class variance method; The fused saliency map is segmented by using the optimal crack segmentation threshold to obtain a candidate crack region image of the fused saliency map. 6. The image detection method according to claim 1, wherein when the crack in the cracked area image is determined to be a real crack, obtaining the crack parameters of the cracked area image includes: respectively calculating the mean value of the gray value of the pixels in the image of the crack area, and the mean value of the gray value of the pixels in a specified area adjacent to the image of the crack area; If the mean value of the gray value of the pixels in the crack area is less than the mean value of the gray value of the pixels in the specified area, then it is determined that the crack in the crack area is a real crack; Binary segmenting the crack region image to obtain a crack binary image of the crack region image, and performing a skeletonization operation on the crack binary image to obtain crack parameters of the crack region image, the crack parameters including crack length and crack width. 7. An image detection device for tunnel cracks, characterized in that the image detection device comprises: The image filtering module is used to perform bilateral filtering processing on the image to be detected of the tunnel crack to obtain the filtered image; a saliency map building module, configured to construct a luminance saliency map of the filtered image and a texture saliency map of the filtered image by using a visual saliency model; A saliency map fusion module, configured to fuse the luminance saliency map and the texture saliency map to obtain a fused saliency map; A saliency map segmentation module, configured to segment the fused saliency map through an adaptive threshold algorithm to obtain a crack region image; The crack parameter acquisition module is configured to acquire the crack parameters of the crack area image when it is determined that the crack in the crack area image is a real crack. 8. The image detection device according to claim 7, wherein the image filtering module comprises: A grayscale variance value statistical unit, used to count the grayscale variance value of each pixel in a specified neighborhood in the image to be detected; an average grayscale variance value calculation unit, configured to calculate the average grayscale variance value of the image to be detected through the grayscale variance value of each pixel; A bilateral filter kernel function construction unit, configured to set a grayscale variance parameter according to the average grayscale variance value, and construct a bilateral filter kernel function based on the grayscale variance parameter; The bilateral filter function calculation unit is configured to use the constructed bilateral filter kernel function to perform bilateral filtering on the image to be detected through a convolution template formula, and use the bilaterally filtered image to be detected as the filtered image. 9. The image detection device according to claim 7, wherein the saliency map construction module comprises: a brightness Gaussian pyramid construction unit, configured to construct a brightness Gaussian pyramid of the filtered image according to a visual saliency model, the brightness Gaussian pyramid comprising a predetermined number of layers; The texture Gaussian pyramid construction unit is used to use the Gabor filter function to calculate the Gabor filter image of each layer image of the brightness Gaussian pyramid in the four directions of 0, π/4, π/2, and 3π/4, Obtain the texture Gaussian pyramid of the filtered image; The luminance saliency map construction unit is used to obtain the image of each layer except the bottom layer from high to low according to the number of layers of the luminance Gaussian pyramid as the image to be processed, and perform upsampling on the image to be processed to obtain the resolution and the image to be processed. an image with the same resolution as the image of the next layer adjacent to the image to be processed, and calculate the absolute value of the subtraction operation between the image obtained by the upsampling and the image of the image of the adjacent next layer to obtain the brightness saliency map; The texture saliency map construction unit is used to obtain the image of each layer except the bottom layer from high to low according to the number of layers of the texture Gaussian pyramid as the image to be processed, and perform up-sampling on the image to be processed to obtain the resolution and the An image with the same resolution as the image of the next layer adjacent to the image to be processed, and calculate the absolute value of the subtraction operation between the image obtained by the upsampling and the image of the image of the adjacent next layer to obtain the texture Significant figure. 10. The image detection device according to claim 7, wherein the saliency map fusion module comprises: A saliency map normalization processing unit, configured to separately obtain the maximum pixel value of a pixel in the luminance saliency map as a first pixel value, obtain the maximum pixel value of a pixel in the texture saliency map as a second pixel value, and dividing the pixel value of the pixel point in the luminance saliency map by the first pixel value to obtain a normalized luminance saliency map, and dividing the pixel value of the pixel point in the texture saliency map by the second pixel value Get the normalized texture saliency map; The fused saliency map construction unit is used to perform pixel value calculation of each pixel in the normalized brightness saliency map and the pixel at the same position as the each pixel in the normalized texture saliency map The comparison, and the maximum value obtained by the comparison is used as the pixel value of the pixel at the same position as each pixel in the fused saliency map to obtain the fused saliency map. 11. The image detection device according to claim 7, wherein the saliency map segmentation acquisition module comprises: An optimal crack segmentation threshold calculation unit, configured to calculate the optimal crack segmentation threshold of the fused saliency map through the maximum inter-class variance method; The saliency map segmentation acquisition module uses the optimal crack segmentation threshold to segment the fused saliency map to obtain candidate crack region images of the fused saliency map. 12. The image detection device according to claim 7, wherein the crack parameter acquisition module comprises: a gray value calculation unit, configured to respectively calculate the mean value of the gray value of the pixels in the image of the crack area, and the mean value of the gray value of the pixels in a specified area adjacent to the image of the crack area; A crack authenticity determination unit, configured to determine that the crack in the crack area is a real crack if the average value of the gray value of the pixels in the crack area is smaller than the average value of the gray value of the pixels in the specified area; A crack parameter acquisition unit, configured to binary segment the crack region image to obtain a crack binary image of the crack region image, and perform a skeletonization operation on the crack binary image to obtain crack parameters of the crack region image, The above crack parameters include crack length and crack width.