KR101667877B1 - Feature detecting method for fish image and fish identification method using feature of fish image - Google Patents

Abstract

A method for extracting feature information of a fish image includes the steps of: receiving a source image including an image of a fish by an image processing apparatus; determining an analysis region in the fish image by the image processing apparatus; Extracting at least one of a feature, a texture feature and a shape feature as feature information of the fish image.

Description

TECHNICAL FIELD [0001] The present invention relates to a method for extracting characteristic information of a fish image, and a method for identifying a fish using characteristic information of a fish image.

The technique described below relates to a method for extracting feature information from a fish image.

Recently, as the importance and concern of ecotourism and ecosystem conservation have increased in many fields, the curiosity about nature is increasing. Due to the development of mobile technology, offline plant inscriptions, insect inscriptions, mushroom inscriptions, and fish inscriptions are available through mobile e-books or related applications.

Korean Patent Publication No. 10-2004-0066777

It is not easy to distinguish a specific region according to the direction of the fish in the fish image in the fish feature extraction or the fish identification using the actual fish image. Particularly in live fish, it is difficult to acquire accurate head or tail regions due to activity. Furthermore, the tail region is smaller than other regions and has a transparent characteristic, making it difficult to distinguish it from the background.

The technique described below proposes an area for extracting fish features from a fish image. The techniques described below are intended to provide a technique for extracting fish feature information from at least one of a shape feature, a color feature, and a texture feature based on an entire fish image and / or a specific region.

The solutions to the technical problems described below are not limited to those mentioned above, and other solutions not mentioned can be clearly understood by those skilled in the art from the following description.

A method for extracting feature information of a fish image includes the steps of: receiving a source image including an image of a fish by an image processing apparatus; determining an analysis region in the fish image by the image processing apparatus; Extracting at least one of a feature, a texture feature and a shape feature as feature information of the fish image.

A method for identifying a fish using characteristic information of a fish image includes the steps of: receiving a source image including a fish image by an image processing apparatus; determining an analysis region in the fish image by the image processing apparatus;

Wherein the image processing apparatus extracts at least one of the analysis region color feature, texture feature, and shape feature as feature information of the fish image, and the image processing apparatus compares the extracted feature information with previously stored standard feature information And determining a fish species of the fish image.

The technique described below distinguishes regions for extracting fish features, and provides feature information that facilitates classification of fish by separately using the types of feature information corresponding to each region.

The effects of the techniques described below are not limited to those mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the following description.

1 is a block diagram showing a configuration of an apparatus for extracting fish feature information.
2 is a block diagram showing a configuration of a system for extracting fish feature information.
3 is an example of a flowchart of a feature information extraction method of a fish image.
4 is an example of a flowchart of a fish identification method using feature information of a fish image.
5 is a flowchart of a process of determining an analysis region in a fish image.
6 is an example showing an analysis region in a fish image.
7 shows an example of generating an GLCM.

The following description is intended to illustrate and describe specific embodiments in the drawings, since various changes may be made and the embodiments may have various embodiments. However, it should be understood that the following description does not limit the specific embodiments, but includes all changes, equivalents, and alternatives falling within the spirit and scope of the following description.

The terms first, second, A, B, etc., may be used to describe various components, but the components are not limited by the terms, but may be used to distinguish one component from another . For example, without departing from the scope of the following description, the first component may be referred to as a second component, and similarly, the second component may also be referred to as a first component. And / or < / RTI > includes any combination of a plurality of related listed items or any of a plurality of related listed items.

As used herein, the singular " include " should be understood to include a plurality of representations unless the context clearly dictates otherwise, and the terms "comprises & Or combinations thereof, and does not preclude the presence or addition of one or more other features, integers, steps, components, components, or combinations thereof.

Before describing the drawings in detail, it is to be clarified that the division of constituent parts in this specification is merely a division by main functions of each constituent part. That is, two or more constituent parts to be described below may be combined into one constituent part, or one constituent part may be divided into two or more functions according to functions that are more subdivided. In addition, each of the constituent units described below may additionally perform some or all of the functions of other constituent units in addition to the main functions of the constituent units themselves, and that some of the main functions, And may be carried out in a dedicated manner. Therefore, the existence of each component described in the present specification should be interpreted as a function. For this reason, the configuration of the components according to the feature information extraction method of the fish image described below achieves the object of the following description It should be clearly understood that the drawings may be different from the corresponding drawings in the limit.

Also, in performing a method or an operation method, each of the processes constituting the method may take place differently from the stated order unless clearly specified in the context. That is, each process may occur in the same order as described, may be performed substantially concurrently, or may be performed in the opposite order.

Hereinafter, a feature information extraction method of a fish image will be described in detail with reference to the drawings.

First, a configuration of an apparatus or system for extracting feature information of a fish image will be described. 1 is a block diagram showing a configuration of an apparatus 100 for extracting fish feature information.

The apparatus 100 of FIG. 1 includes a camera 110 for obtaining a source image including a fish image, a central processing unit 130 for detecting a fish image (fish object) in the source image, And a memory 120 for storing the image obtained by the camera 110 or storing the feature information extracted by the central processing unit 130. The memory 120 may be replaced with a separate storage medium such as a USB or the like.

Meanwhile, the memory 120 may store a program (algorithm) for detecting a fish image in a source image and extracting characteristic information from a fish image. In this case, the central processing unit 130 can detect the fish image through the program stored in the memory 120, and extract the feature information from the fish image.

In addition, the apparatus 100 of FIG. 1 may include a display device 140 for outputting a fish image, an analysis region selected for analysis in the fish image, and feature information extracted from the analysis region. Furthermore, the apparatus 100 of FIG. 1 may further include a communication module 150 for transmitting the fish characteristic information extracted by the central processing unit 130 to another external device. The external device may be a device connected to the Internet, a device such as a smart phone, an AP in a mobile communication network, a PGW in a mobile communication network, a router connected to the Internet, and the like.

2 is a block diagram showing a configuration of a system 200 for extracting fish feature information. The system 200 of FIG. 2 includes a camera device 210 for obtaining a source image including a fish image and a server 250 for extracting feature information from the fish image.

The camera device 210 includes an image sensor 211 for acquiring a source image including a fish image, a communication module 213 for transmitting a source image, a central processing unit 212 for controlling source image acquisition and source image transmission . Furthermore, the camera device 210 may further include a display device 214 for outputting the acquired image. The camera device 210 may be a device such as a camera, a smart phone, or the like having a communication function.

The server 250 includes a communication module 251 for receiving a source image from the camera device 210, an object detection module 252 for detecting a fish image (object) from the source image, an analysis area A specific module 253, and a feature information extraction module 254 for extracting feature information from the analysis area. Further, the server 250 may include a fish identification module 255 for comparing the extracted feature information with standard feature information prepared in advance and determining a fish type or the like. Although not shown in FIG. 2, standard feature information is stored in a separate memory device or the like.

Hereinafter, a method of extracting fish feature information from the apparatus 100 to the system 200 and a fish identification method using the feature information will be described.

3 is an example of a flowchart of a feature information extraction method 300 of a fish image. A method 300 for extracting feature information of a fish image includes the steps of receiving (310) a source image including an image of a fish by an image processing apparatus, determining 320 an analysis region in a fish image, (330) extracting at least one of a color feature, a texture feature, and a shape feature of the analysis region as feature information of a fish image.

4 is an example of a flowchart for a fish identification method 400 using feature information of a fish image. A method 400 for identifying a fish using feature information of a fish image includes a step 410 of receiving a source image including a fish image by the image processing apparatus, a step 420 of determining an analysis region in a fish image, (430) of extracting at least one of a color feature, a texture feature, and a shape feature of the analysis region as feature information of a fish image, and comparing the extracted feature information with previously stored standard feature information, And a step 440 of determining the type.

The fish identification method 400 using the feature information of the fish image extracts the feature information by the same process as the feature information method 300 of FIG. 3 and determines 440 using the extracted feature information . The step 440 of determining the type of fish determines the feature information most similar to the currently extracted feature information among the standard feature information provided in advance and specifies the fish of the current fish as the fish type corresponding to the determined standard feature information. Therefore, the standard feature information is prepared by applying the feature information extraction method 300 to various fish species in advance. Standard feature information may be stored in a memory in the image processing apparatus, a separate external storage medium, or a separate server connected to the network.

The image processing apparatuses of FIGS. 3 to 4 correspond to the apparatus 100 for extracting the fish feature information of FIG. 1 described above or the system 200 for extracting the fish feature information of FIG.

The process of extracting the fish feature information from the fish feature image information extraction method 300 of FIG. 3 and the fish identification method 400 using the feature information of the fish image of FIG. 4 is the same. Therefore, the process of extracting the fish feature information will be described in detail with reference to FIG.

5 is a flowchart of a process 500 for determining an analysis region in a fish image. 6 is an example showing an analysis region in a fish image.

First, the image processing apparatus receives a source image including a fish image (310). In step 320 of determining an analysis region, the image processing apparatus first detects a fish image corresponding to a fish object in the source image (510). The fish image detection 510 may use various techniques for detecting a specific object by distinguishing a background from a specific object.

The feature information may be extracted from the entirety of the detected fish image, but a specific region on the morphological feature of the fish may be used as an analysis region for extracting the feature information. The analysis area is obtained by first obtaining the minimum bounding rectangle containing all the fish objects in the source image. In Fig. 6, the outline rectangle including the fish object corresponds to the minimum boundary rectangle.

The fish object can be divided into five regions based on the minimum bounding rectangle (520). The five segments of the fish object image are divided into three at a ratio of 1: 2: 1 in the vertical direction and three at the ratio of 1: 1: 1 in the horizontal direction. For convenience of description, the head region L1, the body regions L2, L3, and L4, and the tail region L5 are divided as shown in FIG. Referring to FIG. 6, it can be seen that L4 in the body region is a double region, but it is difficult to distinguish the double region from the dorsal fin region in the image processing. For example, when the fish of FIG. 6 is inverted up and down, it is difficult to distinguish between the pear region and the dorsal fin region. L4 is not a code indicating a multiplication area. L2 is the upper region above the image of the torso region, L4 is the lower region below the image of the torso region, and L3 is the middle region between L2 and L4.

Since the difference of color information according to the fish is particularly large in the fish area, it is a key area for extracting feature information. Therefore, it is important to specify the area of the ship.

Recognition of the area recognition recognizes the double area of L2 and L4 among the body areas excluding L1, L3 and L5.

A gray scale conversion is performed on the upper region L2 and the lower region L4 of the trunk region and a gray scale average value for each region is calculated 530. Grayscale conversion may be performed on the entire fish image in advance.

(540). If the gray scale average value of L4 is greater than L2 or equal to L2 (Yes), L4 is determined as a doubled area (550). Conversely, when the gray scale average value of L2 is larger than L4 (No), L2 is determined as the double area (560).

Various criteria can be used for the analysis area such as the whole fish, the fish body area, the fish pear area, and the like.

As described above, the image processing apparatus extracts at least one of a color feature, a texture feature, and a shape feature for an analysis region as feature information of a fish image. Hereinafter, feature information extraction for the analysis area will be described.

Different types of feature information can be used depending on the analysis area. For example, the shape feature can be determined based on the entire area of the fish. The shape feature can be based on the roundness of the whole area of the fish. The degree of roundness may be information having a value of 1 as the number of the circle is approximated as shown in Equation 1 below.

In Equation (1), Area represents the area of the fish object, and Perimeter represents the perimeter of the fish object.

The head region and the tail region of the fish object region are disadvantageously difficult to detect due to the activity of the fish in the fish image. In addition, the background is included in the detection of the tail region, because light or light reflects light due to the thin and transparent characteristic of the fish tail region. Therefore, the color feature can be determined based only on the fish body area. Further, the body region can be determined based on each of the regions L2, L3 and L4. It is also possible to decide based on the area of the boat.

The color features include the R value average of the RGB image, the G value average of the RGB image, the B value average of the RGB image, the Y value average of the YCbCr image, the Cb value of the YCbCr image, the Cr value average of the YCbCr image, At least one of an H value average, an S value average of an HSV image, and a V value average of an HSV image may be used. Of course, in order to utilize all the information of different color space, it is necessary to convert a specific image area (RGB, YCbCr or HSV) in a source image, a fish object detected from a source image, or a fish object into a specific color space.

The texture feature can be determined by applying a texture analysis technique to the image. The texture analysis technique is a method that can utilize spatial information and considers the change of neighboring image brightness value. Texture is a spatial distribution of brightness value changes, so it is very useful to analyze the texture characteristics of images if the spatial shape of the brightness value distribution can be calculated quantitatively. Gray-level Co-Occurrence Matrix (GLCM) can be used as a texture analysis technique.

Since the GLCM uses the gray scale value, the gray scale conversion for the analysis area must be preceded. The gray scale represents an image in the gray color of black and white. It is represented by 256 levels of gray-level from 0 to 255 based on the intensity of light.

7 shows an example of generating an GLCM. FIG. 7A shows an example of generating a GLCM for an image having 4 × 4 pixels. It is assumed that each pixel has four kinds of pixel values from 0 to 3. The left side of FIG. 7 (A) shows pixel values for 4 × 4 pixels, and the right side shows GLCM having 0 to 3 levels. If the left and right neighboring pixels have pixel values of 0 and 0 respectively, the count of the matrix i = 0, j = 0 is incremented by one. For example, in FIG. 7 (A), there are two cases in which pixel values are 0,0 in left and right, three cases in which the left and right pixel values are 2,2, and one case in which the left and right pixel values are 3,3. In fact, the criteria for generating GLCMs can vary. Not only the left and right pixels but also adjacent pixels may be considered, or neighboring pixels that are several spaces apart may be considered.

Further, the normalization process can be performed on the generated GLCM. Equation 2 below is an example of normalization for GLCM.

In Equation (2), V _{i, j} is a value corresponding to the (i, j) coordinate of the GLCM. FIG. 7B shows an example in which the GLCM generated in FIG. 7A is normalized.

A texture feature expressed by the following Equation 3 can be generated based on the normalized GLCM. Contrast, Entropy, Homogeneity, Correlation are used in order of texture features.

In Equation (3), P _{i, j} is a matrix coordinate value generated by GLCM, and (i, j) represents a GLCM value corresponding to a coordinate. μ _i , μ _j is the average for rows and columns, and σ _i , σ _j is the standard deviation for rows and columns.

The shape information may be used for the entire region of the fish when the feature information is summarized, and the color feature and texture feature may be used for the entire body region or for a part of the body region (e.g., a fold region).

It should be noted that the present embodiment and the drawings attached hereto are only a part of the technical idea included in the above-described technology, and those skilled in the art will readily understand the technical ideas included in the above- It is to be understood that both variations and specific embodiments which can be deduced are included in the scope of the above-mentioned technical scope.

100: a device for extracting fish feature information
110: camera 120: memory (storage medium)
130: central processing unit 140: display device
150: Communication module
200: System for extracting fish feature information
210: camera device 211: image sensor
212: central processing unit 213: communication module
214: Display device
250: server 251: communication module
252: object detection module 253: analysis area specific module
254: Feature information extraction module 255: Fish identification module

Claims (16)

Receiving a source image including a fish image by an image processing apparatus;
The image processing device determining an analysis region in the fish image; And
Extracting at least one of a color feature, a texture feature, and a shape feature of the analysis region as feature information of the fish image,
Wherein the determining comprises: converting the source image to a gray scale; Dividing the image of the fish converted into grayscale into 1: 2: 1 in the fish length direction, and dividing into a head region, a body region and a tail region; Dividing the body region into an upper region, an intermediate region, and a lower region by dividing the body region into three equal parts in a height direction of the body region; And determining an area having a larger average value of grayscale among the upper area and the lower area as a double area. The method according to claim 1,
Wherein the analysis region includes at least one of an entire region, a head region, a tail region, and a body region, and the body region includes at least one of three regions divided in the body height direction. The method according to claim 1,
The color feature may include a R average value in the RGB image of the analysis region, a G average value in the RGB image, a B average value in the RGB image, a Y average value in the YCbCr image of the analysis region, a Cb average value in the YCbCr image, Wherein the HSV image has at least one of an average value, an H average value in the HSV image of the analysis region, an S average value in the HSV image, and a V average value in the HSV image. The method according to claim 1,
Wherein the shape feature is a roundness of the entire fish image. The method according to claim 1,
Wherein the texture feature is determined using a GLCM (Gray Level Co-Occurrence Matrix) for the fish image. delete The method according to claim 1,
In the extracting step
Wherein the image processing apparatus extracts at least one of a color feature and a texture feature of the multiplication area as the feature information. The method according to claim 1,
In the extracting step
Wherein the image processing apparatus extracts a color feature and a texture feature of at least one of the upper region, the middle region, and the lower region with the feature information. The method according to claim 1,
Wherein the analysis region is a body region of the fish image,
Wherein the feature information includes at least one of a roundness of the entire fish image, a color feature of the body region, and a texture feature of the body region. 10. The method of claim 9,
The color feature may include an R average value in the RGB image of the body region, a G average value in the RGB image, a B average value in the RGB image, a Y average value in the YCbCr image of the body region, a Cb average value in the YCbCr image, Wherein the HSV image has at least one of an average value, an H average value in the HSV image of the body region, an S average value in the HSV image, and a V average value in the HSV image. 10. The method of claim 9,
Wherein the texture feature is determined using GLCM (Gray Level Co-Occurrence Matrix) for the body region. Receiving a source image including a fish image by an image processing apparatus;
The image processing device determining an analysis region in the fish image;
Extracting at least one of a color feature, a texture feature, and a shape feature of the analysis region as feature information of the fish image; And
Comparing the feature information with standard feature information previously stored in the image processing apparatus to determine a fish type of the fish image,
Wherein the determining of the analysis region comprises: converting the source image to a gray scale; Dividing the image of the fish converted into grayscale into 1: 2: 1 in the fish length direction, and dividing into a head region, a body region and a tail region; Dividing the body region into an upper region, an intermediate region, and a lower region by dividing the body region into three equal parts in a height direction of the body region; And determining an area having a larger average value of grayscales among the upper area and the lower area as a double area. delete 13. The method of claim 12,
Wherein the feature information includes at least one of a roundness of the entire fish image, a color feature of the intermediate region, a texture feature of the intermediate region, a color feature of the multiplication region, A method of identifying a fish using characteristic information of a fish. 13. The method of claim 12,
The color feature may include a R average value in the RGB image of the analysis region, a G average value in the RGB image, a B average value in the RGB image, a Y average value in the YCbCr image of the analysis region, a Cb average value in the YCbCr image, Wherein the HSV image has at least one of a mean value, an H average value in the HSV image of the analysis region, an S average value in the HSV image, and a V average value in the HSV image. 13. The method of claim 12,
Wherein the texture feature is a GLCM (Gray Level Co-Occurrence Matrix) for the fish image.

Images