DE10230200A1 - Method and device for the detection of objects located on a surface of a transparent element - Google Patents

Abstract

The invention relates to a method for the detection of objects located on a surface (9a) of a transparent element (9) with a detector (6) focused at least approximately on the surface (9a) with a plurality of detector fields for detecting a predetermined section on the surface (9a ), whereby a contrast distribution of an image of at least part of the section is used for an evaluation with regard to objects located on the surface (9a). A device for detecting objects located on a surface (9a) of a transparent element (9) is also proposed. The device comprises a detector (6) for detecting a predetermined section on the surface (9a) and an electronics unit (7) for evaluating signals from the detector (6). The electronic unit (7) is designed to be able to evaluate the corresponding surface with respect to objects located on the basis of the contrast distribution of an image of at least part of the section.

Description

The invention relates to a method and a device for detecting on a surface of a transparent elements located objects according to the generic term of claims 1 and 5.

State of the art

The detection of on the surface of a transparent Elements located objects are known for example by Realized rain sensors with which the presence of visual obstructions monitored on a windshield to control the wiper system of a vehicle.

Most of the rain sensors used today work according to the reflection principle. Such a sensor is, for example, in the German published application DE 199 43 887 A1 disclosed. This sensor takes advantage of the physical effect that light rays falling on the windshield from the inside experience total reflection on the outer surface of a clean windshield. The intensity of the reflected rays is measured with a photodetector. If there are disturbing objects, such as raindrops or dust on the windshield, some light rays are decoupled or scattered. The amount of light arriving at the photodetector is reduced in this case, which can be used for a corresponding control of a wiper system.

A rain sensor based on the reflection principle however, there can be different types of visual impairments do not differentiate. For example, the rain sensor reacts both on dust as well as on raindrops in the same way. This leads to circumstances to an undesirable Behavior of a wiper system. For example, the wiper system in the case of dry soiling without the addition of cleaning water Gear set, causing wiper blades squeak or, if it is a sand film, even one Scratching the windshield occurs.

One way to solve this problem was in the German published application DE 197 49 331 A1 described. In this embodiment, instead of simple photo detectors, an optical sensor array is used, which is focused on the outside of the windshield. Images recorded with the sensor array are Fourier transformed. The spatial spectra obtained from this are compared with reference spectra of certain types of pollution. This enables a statement to be made about different types of contamination.

Task and Advantages of the invention

The invention has for its object a alternative possibility to provide, in particular, an assessment with regard to carry out certain types of contamination of a windshield to be able to.

This task is due to the characteristics of claims 1 and 5 solved. In the dependent claims are advantageous and expedient further training described the invention.

The invention starts from a method for the detection of on a surface of a transparent Elements, e.g. objects located on a windshield out. For this purpose, a detector that is at least approximately focused on the surface is used with multiple detector fields, especially a video sensor array used, with which a predetermined section on the surface can be recorded. The essence of the invention is now that the contrast distribution an image of at least a portion of the detail for evaluation on the surface existing objects is used. This approach lies based on the knowledge that this was detected by the detector on the surface Image in a comparatively small image area strong contrasts shows when on the surface Objects, e.g. Water drops or dust particles are from the to the surface focused detector can be mapped accordingly. are no objects on the surface to be detected the detector is just a blurred, blurred image of the surroundings, on which he is not focused. This blurry Image is especially when viewed comparatively on a small scale in contrast. The type of contrast distribution is specific in many cases for the Type of pollution. For example, a learning machine can be on it be trained to take pictures with sharply depicted objects small-scale Consideration shows strong contrasts, from those with blurred objects to distinguish and additionally a differentiation of sharply depicted objects for classification the pollution.

In a further particularly preferred embodiment the invention is based on the contrast distribution of a Image of at least part of the section of the detected Surface one Function or value determines which in a comparison as a measure of an evaluation regarding on the surface objects in use. There are various ways of determining a measure of contrast Processes from digital image processing possible. Ultimately comes it depends on a function or a value for the contrast of an image to generate who or with a known function or can be compared to a known value in order to provide a basis for decision-making to determine whether e.g. the windshield is dirty and, if applicable, what type of contamination is present.

In a further preferred embodiment According to the invention, an image recorded by the detector is broken down into several images with decreasing resolution. This process is also called multiscale analysis. Then a contrast measure is determined for the individual images. From the relationship between contrast measure and resolution, a statement can be made about objects on the surface. The captured image is preferably broken down by repeated application of smoothing operations with suitable filters. A contrast measure, for example the standard deviation of the intensity values of the image points, is determined in each resolution level. If the surface, for example the outside of a windshield, is free of objects, only blurred objects from larger distances are shown. The function contrast measure over resolution therefore shows higher contrasts if necessary only at low resolutions. If, on the other hand, there are objects on the writing surface, fine details are also reproduced, which means that the contrast measure over resolution function shows large contrast values even at comparatively higher resolutions.

An assessment as to whether a sharp or unsharp image is present can also be made by determining how the image points of the image depend on one another with respect to the intensity values and / or the color. In the case of, for example, a blurred image, differences in intensity, in particular between neighboring pixels, are comparatively small. The neighboring pixels are strongly dependent on each other. With a ^sharp picture z. B. sharp differences in intensity between adjacent pixels on sharp edges. This fact can in turn be used for the evaluation of objects on the surface to be detected. In order to determine the interdependency of pixels, an autocorrelation function can be determined by expanding an image. An autocorrelation function for a free surface has a very specific course, which makes it possible to distinguish it from an autocorrelation function for a surface with dirt. In addition, auto-correlation functions can be assigned to specific types of contamination on the surface.

In a device for detection from on the surface a transparent element, e.g. a windshield, yourself disturbing Objects with an at least approximately focused on the surface Detector with several detector fields to capture a given section on the surface and an electronics unit for evaluating signals from the detector the key idea is that the electronics unit is designed for this is, based on the contrast distribution of an image of at least part of the section of the surface with respect to the corresponding surface to be able to evaluate objects on it. The gained from it Information can then z. B. for the control of a wiping position can be used. As already in view explained to the process the detector will only be able to take sharp, high-contrast images, provided Objects on the surface available. Otherwise, that is recorded by the detector Image is blurry and therefore low in contrast. To between sharp and The electronic unit can distinguish blurred images an automatic classifier, e.g. a neural network or comprise a polynomial classifier. Such a facility can be trained on it, not just between sharp and blurred Images, but also various sharp images, i.e. various Differentiate between types of contamination.

be a further preferred embodiment The invention can be used for an evaluation using the electronic unit regarding on the surface objects in a comparison with given information carried out become. For example, for a captured image determines a measure of contrast that with predetermined Contrast measures known Images being compared for which is the type of pollution or whether pollution at all is known. In this context it is further preferred if the electronics unit for disassembling one picked up by the detector Image into multiple images with decreasing resolution and to determine one contrast measurement for the each picture is laid out. The resulting contrast distributions can compare the electronics unit with known contrast distributions, to get information about to get pollution and its nature.

The same applies to the case if the electronics unit is able to determine how pixels an image at a predetermined distance in the plane of the image depend on each other in terms of contrast, e.g. by creating a Autocorrelation function. That is preferably taken from a Image calculates an autocorrelation function and with autocorrelation functions with known surface coverings with objects, e.g. different Soiling a windshield, compared to the resulting currently on the surface to be able to classify existing objects. The resultant The result can then be used to specifically control a wiper system.

In a particularly preferred embodiment According to the invention, the detector is a video sensor, in particular a Video sensor array on which a predetermined one to be detected area can be reproduced. In this context, it is advantageous if between detecting surface and a lens unit is provided to the video sensor. So that can the area to be detected is sharp are mapped onto the sensor array.

With an appropriate resolution of the video sensor array, this structure can be used very well make differentiated statements about the surface and objects on it.

To the device according to the invention to different lighting conditions to be able to adapt will about it also proposed that a module for exposure control of the Detector is present.

A device according to the invention let yourself through the use of means for adjusting the focusing level use the sensor for environmental detection.

Several exemplary embodiments for illustration the invention are illustrated in the drawings and with details further advantages and details explained. It

demonstrate

1a and b the contrast of an image captured by a detector on a pane, plotted against the resolution of the image with a clear view ( 1a ) and in the event that interfering objects are on the window ( 1b )

2a and b Corresponding function profiles with which a differentiation of interference objects is to be illustrated and

3 a highly schematic structure of a video-based rain sensor.

description of the embodiments

In the 1a to 2 B a contrast measure K is shown as a function of the resolution A.

Under contrast or measure of contrast is in the sense of the invention in general one of the difference between bright and dark values (intensity values) the size of the individual pixels of the captured motif, if applicable understood in a weighted form.

In the 1a shown function 1 for the contrast K results from an image of a surface of a pane, which was recorded by a video sensor focused on the surface. The video image of the video sensor is broken down into several images with decreasing resolution by the repeated application of a smoothing operation by means of an electronic unit. A global measure of contrast is calculated at each resolution level, for example the standard deviation of the intensity values of the pixels. The corresponding measure of contrast, plotted against the resolution, results in the function 1 for a disc that is free of objects. In this case, the video sensor only detects unsharp objects from a greater distance. The blurred, blurry images show an increase in contrast only at comparatively low resolutions.

If, on the other hand, there are interfering objects on the pane surface detected by the video sensor, the function caused by the background 1 through a function 2 superimposed, which originates from the interfering objects on the surface of the pane. this function 2 typically shows an increase in contrast even at comparatively higher resolution values, since the video sensor detects the interfering objects on the pane surface as a sharp image with correspondingly small-scale jumps in intensity. These small-scale jumps in intensity lead to a correspondingly high contrast even at high resolutions of the image.

In the 2a and 2 B is the function 1 that is representative of the background by itself from the function 2 distinctive functions 3 or 4 overlaid.

When evaluating a course of contrast over the dissolution one can take advantage of the knowledge that different interference objects are characteristic courses of contrast over the resolution demonstrate.

In 2a for example, represents the function 3 a sand film which, with the appropriate grain size, leads in extreme cases to an isolated tip in the contrast spectrum. However, the function shows 4 in 2 B a broader distribution of contrast, which is characteristic of raindrops.

After an appropriate analysis and evaluation of the images captured by the video sensor cannot just a statement about it be taken that interfering objects on the disc surface but for in the event that these are detected, a classification immediately be made. Dependent on of which can then e.g. a windshield wiper system are used. If raindrops are detected, actuating the wiper blades is sufficient. is on the other hand, if there is a sand film, the splash water position should be used when using the wiper blades with actuated be squeaky noises to avoid scratching the window.

In 3 is a schematic representation of a video-based rain sensor 5 displayed. The rain sensor 5 consists of an optical sensor array 6 to which a data processing unit 7 connected. Via a lens unit 8th becomes the outside 9a a windshield shown schematically here 9 on the sensor array 6 displayed. The rain sensor 5 is preferably behind the window 9 attached and on the outside 9a the windshield 9 focused. One includes the data processing unit for evaluation 7 a module 10 to control the exposure of the sensor array 6 , a measuring unit 11 for contrast measurement and a decision module 12 , The decision module 12 is designed in such a way that it can make an assessment of what is available and, if necessary, of the type of visual impairment.

This information can then be according to the arrow 13 an actuator (not shown) are supplied.

Is preferably between the lens unit 8th and the sensor array 6 an aperture (not shown) is attached to reduce the depth of field of the optics. This allows a better separation of background and foreground contrasts.

From video images that immediately after a wiping process, reference distributions can be made directly e.g. For the contrast over the resolution on the basis of which the next wiping process is initiated is decided. This is a situation-specific adaptation of wiping behavior possible.

In situations where the ambient brightness illuminates the object on the pane 9 not sufficient (e.g. at night), it may make sense to switch on (e.g. IR-based) window lighting. The light reflections caused then again allow, for example, object detection based on the contrast distribution over the resolution.

Corresponding results can be achieve if there are differences in brightness between the pixels can be evaluated by an autocorrelation function.

Also the solution described above with Help a learning machine that recognizes whether a sharp or unsharp one Picture and the one above a differentiation between different types of pollution carry out can is for certain use cases advantageous.

Claims (12)

Method for the detection of on a surface ( 9a ) of a transparent element ( 9 ) objects on the surface ( 9a ) at least approximately focused detector ( 6 ) with several detector fields to detect a given section on the surface ( 9a ), characterized in that a contrast distribution of an image of at least part of the section for an evaluation with regard to the surface ( 9a ) located objects is used. A method according to claim 1, characterized in that on the basis of the contrast distribution of an image of at least part of the section of the surface ( 9a ) a function or a value is determined which is used as a measure for an evaluation with regard to the surface ( 9a ) located objects. Method according to claim 1 or 2, characterized in that an image is broken down into a plurality of images with decreasing resolution and a measure of contrast is determined for the individual pictures, and that a statement about itself on the surface ( 9a } objects found. Method according to one of the preceding claims, characterized characterized that it is determined how pixels of the image in a given distance in the plane of the image with respect the intensity and / or the color depend on each other. Device for the detection of on a surface ( 9a } of a transparent element ( 9 ) objects on the surface ( 9a ) at least approximately focused detector ( 6 ) with several detector fields to detect a given section on the surface ( 9a ) and an electronic unit ( 7 ) for evaluating signals from the detector ( 6 ), characterized in that the electronic unit ( 7 ) is designed to be able to evaluate the corresponding surface with respect to objects located thereon on the basis of the contrast distribution of an image of at least part of the section. Apparatus according to claim 5, characterized in that the electronics unit ( 7 ) is designed such that it can be used for an evaluation with regard to the surface ( 9a ) objects in a comparison with given information. Apparatus according to claim 5 or 6, characterized in that the electronics unit ( 7 ) to disassemble one from the detector ( 6 ) recorded image is designed into several images with decreasing resolution and for determining a measure of contrast for the respective image. Device according to one of claims 5 to 7, characterized in that the electronic unit is able to determine how pixels of an image in a ^v orgegebenen distance in the plane of the image of the contrast dependent with respect to one another, to create in particular an auto-correlation function. Device according to one of claims 5 to 8, characterized in that the detector is a video sensor, in particular a video sensor array ( 6 ) includes. Device according to one of claims 5 to 9, characterized in that between the detector ( 6 ) and the surface to be detected ( 9a ) a lens unit ( 8th ) is provided. Device according to one of claims 5 to 10, characterized in that a module ( 10 ) for exposure control of the detector ( 6 ) is available. Device according to one of claims 5 to 11, characterized in that means for adjusting tion of the focusing plane are provided.