JPH04240556A - Apparatus for measuring crack of road surface - Google Patents

Abstract

PURPOSE:To store the measured data necessary for later analysis in a memory medium of the min. memory capacity during measuring work by mounting a measuring and processing apparatus and performing two-dimensional image processing within a real time along with the position data of a truck. CONSTITUTION:Respective segment images of edge data subjected to characteristic extraction by a series of processings through the smoothing processing part 41, density threshold value determining part 42, semi-threshold value processing part 43 and edge extraction processing part in a measuring and processing apparatus 4 become crack candidates. The areal evaluation of crack images at every image units is performed for each unit in an area evaluating and judging part 46 and the position coordinates values thereof are allotted to the crack image data and, further, the area ratios of the crack candidates are calculated on the basis of the ratio of the total number of the pixels of each label to a unit section area to be compared with a preset threshold value. With respect to the unit section exceeding the threshold value, image data containing position coordinates and respective label coordinates data is outputted to the unit section as recording data. This data is recorded on a recording apparatus 6.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] This invention continuously measures and measures cracks in paved surfaces such as roads and runways using image processing.
The present invention relates to a measuring device for measuring cracks on a road surface, and particularly to a measuring device suitable for obtaining measurement data for evaluating and determining cracks using a pattern recognition method using a personal computer or the like.

0002

[Prior Art] As a method for evaluating the degree of damage to paved surfaces such as existing roads and runways, a width of 1 [m] is used for each unit area of 50 [cm] square for an inspection section of several kilometers.
The presence or absence of cracks of 50 m or more is measured, and the crack rate (%) is calculated from the ratio of the number of cracked surfaces to the total number of measurement unit surfaces, and this is used as a guideline for road surface repair. Conventionally, cracks have been measured for each unit area by visual inspection at the site or by visual judgment from images taken with a video camera or by taking photographs for each unit area, and the main method for data output is sketching.

0003

[Problems to be Solved by the Invention] Conventional methods for detecting cracks in pavement surfaces require a huge amount of time for workers to visually observe, take videos or photographs, visually inspect the images, and create sketches. However, in addition to the huge amount of videotapes and photographic films taken home, different inspectors have different criteria for detecting cracks, making it extremely difficult to detect cracks efficiently using fixed criteria.

[0004] The object of this invention is to efficiently and automatically measure cracks on the surface of the road surface to be inspected by being towed by a car. An object of the present invention is to provide a road surface crack measuring device equipped with a measurement processing device capable of storing information on a storage medium having a storage capacity of .

[0005]

[Means for Solving the Problems] In order to solve the above-mentioned problems, in the road surface crack measuring device according to the present invention, a vehicle is mounted on a cart that is movable on the road surface, and the road surface directly below the road surface is viewed as a field of view. an imaging device that outputs one-dimensional image information of a road surface in a predetermined number of pixels in a predetermined number of pixels; an illumination device having a light source placed above the ground contact surface of the trolley; and a lighting device that captures the one-dimensional image information from the imaging device every predetermined movement distance of the trolley, and captures the one-dimensional image information from the imaging device in real time together with the position information of the trolley. The measurement processing device is equipped with a measurement processing device that performs two-dimensional image processing, and a recording device that records information output from the measurement processing device on a removable recording medium. density threshold determining means for determining a density threshold from an average density value of image pixels for the one-dimensional image information for a predetermined number of scans to measure the average brightness of a road surface; On the other hand, means for performing half-threshold processing for uniformly converting and fixing pixel densities on the high luminance side corresponding to background information with the density threshold as a boundary, and focusing on scanning in the two-dimensional image processing. Edge extraction processing that performs a conversion process on a point pixel using the maximum value of the absolute value of the difference in density between pixels adjacent to the point of interest in four directions (vertical, horizontal, and diagonal) centered on the point of interest as point of interest data. and for each pixel of the image data after the edge extraction processing, the sum of the densities of surrounding pixels that are separated by a predetermined pixel from the point of interest in a total of four directions (vertical, horizontal, and diagonal) with the point of interest as the center is defined as point of interest data. a noise removal processing means that performs a conversion process to perform a conversion process; Area evaluation determination processing means for evaluating and outputting binary image information of only the unit section having a crack candidate area exceeding a predetermined threshold value to the recording device together with the position information of the unit section. .

[0006]

[Function] In the road surface crack measuring device according to the present invention, the imaging device mounted on a trolley that can be moved by towing or the like on the road surface includes an imaging device such as a CCD linear image sensor and an optical system, and One-dimensional image information of the road surface in a direction intersecting the moving direction is outputted with a predetermined number of pixels for each predetermined distance of travel of the truck, with the road surface being viewed as a direct view from the top of the truck. That is, from this imaging device, one-dimensional image information in the width direction of the road surface is obtained every time a signal for each unit movement distance (for example, 1 [mm]) from a rangefinder such as an encoder based on the wheel rotation of a truck arrives. taken out. The number of pixels of this one-dimensional image information corresponds to the number of pixels of the linear image sensor used. For example, if the minimum width of a crack to be measured is 1 [mm], the resolution needs to be about 0.5 [mm], so a field of view width of about 2 [m] in the width direction is divided into two pieces of 2048 pixels. CC of
It is sufficient to cover it with 4096 pixels using a D linear image sensor. In this case, the maximum towing speed of the trolley is 4 [km/km].
h], then the CCD linear image sensor is 2.5[
It is sufficient to drive the video at a video rate of approximately MHz. Also,
The data format of the one-dimensional image information output from the imaging device is digital data in which each pixel is expressed, for example, in 8-bit (256 gradation) gradations.

The illumination device includes a hood that shields the road surface within the field of view from the surroundings, and within the hood there is a space provided between the field of view and the ground plane of the trolley at a distance in front or rearward in the direction of movement of the trolley. A high-intensity light source, such as a rod-shaped halogen lamp for a copying machine, is placed above the lamp and extends along the intersecting direction, so that the field of view can be illuminated at a low illumination angle. In this way, in the present invention, the imaging area is covered with a hood and the illumination device illuminates the road surface at a low illumination angle, and the imaging device captures images in the normal direction to the road surface illuminated at a low angle within the hood. The optical system is arranged to obtain an image from directly above, and the dark shadow created by the edge of the crack is captured while avoiding the influence of external light. Conditions such as the position of the sheath illumination light source and the irradiation angle are preferably calculated from theoretical calculations and average shape data of the actual road surface, and confirmed and determined through experiments.

For example, on a road with a width of 3.1 [m], the influence of the tilt of the bogie due to the presence of ruts for various vehicles can be significantly reduced by setting the wheel width of the bogie to 3.0 [m], which is close to the road width. It has been confirmed that it can be reduced. Therefore, the width of the road to be measured is 2.0, 2.5, and 3.0, respectively.
, 3.5 [m], the wheel width of the cart is 1.9, 2.
It is preferable to set the distances to 4, 2.9, and 3.4 [m], respectively, so that both the left and right wheels of the truck roll on the outside of the left and right ruts on the road surface. When moving a truck with such a wheel width, it is an important consideration in setting the conditions of the imaging device and lighting device that the height of the road surface changes while taking into account the meandering of the truck. For example, on a road that was actually studied, if it is assumed that the meandering of the bogie swings left and right by ±100 [mm], the vertical positional fluctuation of the road surface with the contact surface of the bogie wheels as the reference plane is at most upward within the road width. 20 [mm]
It is about 50 [mm] downward.

In this case, the illumination device should not interfere with the road surface, and the illumination light source should be brought as close to the road surface as possible so that the shadows at the edges of the cracks described above can be clearly produced even when there are fluctuations in the road surface height. In order to achieve this, it is preferable to set the height of the illumination light source from the trolley ground reference plane to about 70 [mm] with a margin of about 50 [mm]. In this case, the horizontal distance from the imaging device's reading line to the illumination light source is also
It is preferable to set the angle to about 70 [mm] so that the irradiation angle of the illumination light is 60 degrees or less at the worst point in the downward direction of the vertical fluctuation of the road surface position. As a result, the irradiation angle of the illumination light with respect to the reference plane becomes a low angle of about 45 degrees.

The brightness of the illumination light source affects the measurement speed (truck traction speed) in relation to the sensitivity of the imaging device, but when using a CCD line image sensor, the brightness of the illumination light source is sufficient as the sensor output voltage within the range of the above-mentioned vertical fluctuations of the road surface. It is only necessary to design it in relation to the sensitivity of the CCD and the storage time so that a suitable size can be obtained. Note that the sensitivity of the CCD in this case is the ratio between the saturation output voltage and the saturation exposure amount of the CCD.

[0011] The measurement processing device first takes in the one-dimensional image information obtained and performs real-time two-dimensional image processing together with the position information of the trolley, and processes the image information for each road surface unit section of a predetermined area. The crack candidate area is evaluated from the number of pixels of the crack candidate, and then binary image information of only the unit section having a crack candidate area exceeding a predetermined threshold is output together with the position information of the unit section. That is,
The one-dimensional image information output from the imaging device becomes two-dimensional image data obtained by scanning the road surface in a raster pattern with a predetermined number of horizontal scanning line pixels as the trolley moves. Real-time processing such as edge detection using a 3×3 pixel matrix operator is performed to convert binary data for recording.

For example, in the two-dimensional image information based on the one-dimensional image information from the imaging device, the luminance (density) of the cracked portion is lower than that of the background portion;
This is relatively low, and crack images may not be detected successfully even if the binarization process is simply performed. In such a case, it is preferable to compress the two-dimensional image information by first binarizing the data by half-width processing using a predetermined density threshold. A practical method for determining this concentration threshold is to measure the average concentration of the target road surface prior to measurement. In this case, the one-dimensional image information for the full width of the imaging field from the imaging device may be averaged for, for example, 128 consecutive lines (or sampling every multiple lines among them), and this may be used as the density threshold. . This half-width processing using the road surface average density as a threshold can be performed by, for example, treating the observed image as 7 bits (128 levels) and treating the observed data above the density threshold as a constant brightness value. can be compressed into 6 bits (64 levels) of information. This realizes processing more suitable for real-time processing as the cart moves.

[0013] Next, the edges of the cracks are detected from the compressed image information and binarized. For this, the edge detection method in the data conversion process can be used. Generally, for example, the central pixel of a weighted matrix of 3 x 3 pixels is taken as the point of interest pixel, and the value obtained by multiplying the density of the surrounding pixels by a predetermined weighting coefficient and adding the resultant data is the converted data of the point of interest data. Edge enhancement methods such as the so-called Sobel operator are often used, but in order to take advantage of these effects and perform real-time processing, a simpler operator in terms of hardware can be defined as follows to enhance edge enhancement. Perform detection processing.

That is, when the data at the point of interest is f(x, y) and the converted data is g(x, y), g(x, y)=Max{h0(x, y), h1(x ,y
), h2 (x, y), h3 (x, y)} h0 (x, y
)=|f(x,y-1)-f(x,y+1)|h1(
x,y)=|f(x+1,y-1)-f(x-1,y+
1) | h2(x,y)=|f(x-1,y)-f(x
+1,y) | h3(x,y)=|f(x-1,y-1
) - f (x+1, y+1) | This operator takes the absolute value of the difference in density between pixels adjacent to the point of interest in a total of four directions (vertical, horizontal, and diagonal) centered on the point of interest, and focuses on the maximum value among them. This is the process of converting it into point data.

The image data obtained in this way is difference value image data with the luminance difference as a scale, and if a crack exists, the value at the boundary becomes high. This difference value data is
Furthermore, focusing on the presence or absence of edge features, noise processing is performed using an appropriate threshold value and binarization is performed. This threshold value is determined by how much difference in brightness is expected between the road surface and the cracks, and it is necessary to set it so that the boundaries of the cracks do not disappear as much as possible, but in this case there is still the risk of picking up noise. Therefore, in noise removal, noise that can be determined by looking at the image locally is a problem, and especially for real-time processing, for example, 3
It is preferable to judge a size within a certain area, such as x3 pixels, to be noise, and to perform conditional judgment using the following operator, which can be simply processed on hardware.

That is, when the data at the point of interest is f(x, y) and the converted data is g(x, y), g(x, y)
=f(x-2,y-2)+f(x,y-2)+f(x+
2,y-2)+f(x-2,y)+f(x+2,y)+
f(x-2,y+2)+f(x,y+2)+f(x+2
,y+2) as an operator, and when the value of this operator is 0, the converted data g(x,y)=
This is the process of setting it to 0 (see FIG. 9). In this way, the data of the calculation results are saved for each pixel of interest, and a series of multivalued data corresponding to the degree of change in density of the outline edge of the grayscale image is generated.

Next, the area of a portion that appears to be a crack (crack candidate) is evaluated from each of the binarized edge image data, and a unit block containing the crack candidate is extracted. In real-time processing, noise and cracks are distinguished (determined) on an image-by-image basis, and features are extracted from images as continuous line figures as cracks, for example, based on two-dimensional dimensions exceeding a predetermined value or degree of directional convergence. Therefore, in the present invention, such analysis processing is carried out by taking the measurement data home and leaving it to processing by a personal computer or the like. Therefore, in the measurement processing device of the crack measuring device of the present invention, in order to efficiently record the crack measurement data (crack position and crack shape in two-dimensional coordinates on the road surface) after the noise processing into the recording device, additional data is required. Perform compression. In other words, a method often used to record this type of image data is to record it on videotape, but this method has insufficient resolution, making it difficult to record in a way that allows playback of even small cracks. be. Furthermore, since the reproduced image on the tape is intended for human viewing, automatic analysis using a personal computer or the like requires further A/D conversion, which is unreasonable in terms of accuracy.

The present invention adopts a method of recording digital data on a recording medium, and in that case, further compresses the data so that only the minimum necessary data is recorded, thereby creating a commercially available digital audio tape (DA).
T) etc. can be used as a recording medium. In other words, it is possible to digitize and record the luminance information for all of the crack measurement data extracted as described above, but in this way, for example, the road surface will be 0.
．． If it is carved into 5 x 0.5 [mm] meshes and each mesh corresponds to one pixel, at an imaging field of view width of approximately 2 [m], each measurement length of 1 [mm] requires 2 kbytes, and at 1 [m]. 2 MB, 1 [km] requires a capacity of 2 GB, which requires either increasing the capacity of the recording device or installing a tape autochanger, and the amount of data processed and the amount of data required by the later analysis computer. It becomes enormous and requires a large-scale computer and memory.

Therefore, in the present invention, the crack candidates detected as described above are evaluated for each unit section of a predetermined area of the road surface. This can be done, for example, on the road surface mentioned above.
In the example of carving into 0.5×0.5 [mm] meshes and making each mesh one pixel, the unit section is 128×
Divide into blocks equivalent to 128 pixels, and for each block, set a plurality of consecutive pixels corresponding to a crack with a predetermined width, for example, 1 [mm] or more in width, as "1", and background information as "0",
A method is adopted in which a section having a predetermined number of consecutive pixels of "1" is detected, and coordinate data of the unit section is recorded together with binary image data including the mesh position of the "1" pixel. For unit sections that do not include such crack candidate data, the measurement processing device does not provide output to the recording device. As a result, the original information of 1 bit per pixel can be compressed to 128 x 128 bits (2 kbytes) per unit section. Such data compression makes it possible to record data in real time as the trolley moves.

[0020] The recording device records the compressed data thus outputted from the measurement processing device onto a recording medium such as a digital audio tape (DAT). These are position coordinates and binary image information within the unit section. After taking the recording medium home, the data recorded on this recording medium is subjected to analysis processing using a personal computer (PC), etc., and through this analysis processing, for example, the characteristics of the shape of cracks in each unit section can be extracted, and the crack rate (i.e., Calculation of the total area of meshes with data "1" (in some cases, the percentage of the road surface repair area added to this) with respect to the section area, and output of a sketch diagram of cracks are performed.

[0021]

[Example] In order to further understand the features and advantages of the present invention, an example applied to a crack measuring device for a paved road surface will be described below with reference to the drawings. This is a schematic perspective view showing the general configuration of the measuring device, in which a trolley 1 that can run on wheels 5 also serves as a box-shaped hood that opens downward, and is equipped with a traction rod 7 at the front for towing by a car. Another truck (power truck) 10 carrying a power supply device with a generator is connected to the rear of the vehicle.

On the trolley 1, there is a camera unit 8 housing a camera 3 (FIGS. 2 and 4) having a CCD linear image sensor as an image sensor and its optical system (not shown), and a measurement processing device 4 (FIG. 4). ) and a measurement unit 9 housing a DAT recorder 6 (FIG. 4). Inside the hood of the trolley 1, as shown in FIGS. 2 and 3, there are two rod-shaped halogen lamps 2 as light sources in the front and two in the rear, with the imaging line H by the camera 3 in between. They are placed alternately at intervals. The camera 3 captures the road surface R with its field of view facing directly below, and its imaging line H faces in the road width direction intersecting the direction of movement of the trolley 1 by the wheels 5. The halogen lamp 2 used as a light source has a total length of 497 [mm], an effective light source length of 448 [mm], an aperture of 6 [mm], a power consumption of 260 [W], and a total luminous flux of 4400 [mm].
lm], efficiency 16.91 [lm/W], the horizontal distance between the lamp and the imaging line is 70 [mm], and the lamp height is 70 [mm] from the wheel ground reference plane of the trolley. The irradiation angle of the illumination light is set to 45 degrees in the reference state. The camera 3 has two 1024-pixel CCD linear image sensors arranged vertically to capture a field of view in the width direction of 2 [m] with a total of 2048 pixels.

The hood of the truck 1 covers the imaging field of view of the road surface R from external light, holds a halogen lamp 2 inside, and illuminates the imaging line H on the road surface with high brightness from a low angle of 45 degrees. The imaging camera 3 constituting the imaging device has a total of 2048 pixels of 8 bits (2
It obtains one-dimensional image information of a digital grayscale image (56 tones), and is provided in a fixed positional relationship with the hood 1 so as to image the road surface from directly above. When an imaging field of the road surface is imaged by the imaging section having such a configuration, the road surface portions other than the cracks become highly bright due to the low-angle illumination light, and the cracks are emphasized by the strong shadows of their edges, so that the imaging camera 3 In the digital grayscale image obtained from , pixels corresponding to strong shadow areas such as cracks are dark and low density, and background areas other than cracks are bright, uniform and high density, and the relative density of the two is The difference becomes greatly emphasized.

FIG. 4 is a block diagram showing the configuration of the signal processing system of the device of this embodiment. The measurement processing device 4 receives one-dimensional image information from the camera 3 through an encoder (not shown) driven by the wheels 5 of the trolley 1, etc. The sensor captures pulse signals from the rangefinder at fixed distance intervals and processes them as two-dimensional image information. This measurement processing device 4 includes a smoothing processing section 41 for reducing noise during observation on the two-dimensional image information.
, a density threshold determination unit 42 that calculates a density threshold M from the average density value of image pixels for the one-dimensional image information for a predetermined number of scans in order to measure the average brightness of the measurement road surface; A half-threshold processing unit 43 performs half-threshold processing by converting the density of two-dimensional image information after smoothing using a threshold value M, and the aforementioned matrix operator is used for the two-dimensional image information after half-threshold processing. An edge extraction processing unit 44 generates a series of difference value image data corresponding to the degree of density change of the contour edge of the grayscale image by performing data conversion processing; Noise removal binarization processing unit 4 that binarizes and outputs binarized edge image data by
5, and an area evaluation determination 46 for evaluating the area of the edge image data for each unit section. In this embodiment, the measurement processing device 4 includes the illustrated functional elements, which are configured by, for example, a microcomputer.

In the half-threshold processing, the density threshold M is applied to the two-dimensional image information prior to the edge extraction processing.
The pixel density on the high luminance side corresponding to the background information is uniformly converted and fixed to the same value as the density threshold value. This improves the binarization accuracy in data conversion processing for subsequent edge extraction processing, and also performs information compression to achieve real-time processing.

The area evaluation determination unit 46 processes the area evaluation process for each predetermined unit section of the binary image data from which noise has been removed, and calculates the area of the unit section and the position coordinates of the unit section only for the unit section that includes a crack candidate having a predetermined area or more. The digitized image information is output to the DAT recorder 6 as a recording device. Each line segment in the image data after feature extraction is a crack candidate, and the determination as to whether it is a crack or not is made on a computer in another location, but here, the area evaluation determination unit 46 analyzes the measurement data on site as much as possible. Processing is performed to compress the data into a small amount and record it on the DAT recorder 6.

That is, the two-dimensional image data from the noise removal binarization processing section 45 is two-dimensional data having a width of 2048 pixels corresponding to the imaging field width 2 [m],
The area evaluation determination unit 46 divides this into 16 blocks each having a unit area of 128×128 pixels and processes the area. The area evaluation determination unit 46 compares the total area of crack candidates for each unit section with a preset threshold, and calculates the position coordinates of the unit section and binary image data only for the unit section whose crack candidate area is equal to or larger than this threshold. Pass it to DAT recorder 6. The DAT recorder 6 digitally records the digital data output from the area evaluation determination section 46 on the DAT, but in this case, it is normal circuit design to send a warning of the remaining amount of tape to the driver's seat of the towing vehicle. It is possible, and it is also possible to obtain monitor output from each necessary part of the measurement processing device.

The crack measurement operation by the measurement processing device 4 in the crack measurement device having the configuration shown in FIG. 4 will be explained below with reference to FIGS. This is two-dimensional image information of an 8-bit (256 tones) digital grayscale image with a width of 2048 pixels captured from the camera 3 as the lighting lamp 2 moves.
Background noise such as shading and color unevenness of the road surface R is reduced in advance by making it uniform brightness using low-angle illumination, and also by focusing on the edge part of the crack and making the shadow stronger, the edge part Before detection, the density of the crack boundary area is emphasized in the image information compared to the non-cracks area of the road surface.

Image processing device 4 that handles such input images
The flow of the image conversion algorithm is shown in FIG. In other words, the input 2048 pixel width 8 bits (
The two-dimensional digital image information (256 tones) is first smoothed by the smoothing processing unit 41 using the matrix operator shown in FIG. 6, and isolated changes in brightness of pixels are smoothed. Although this input image information has been density-enhanced, it still contains noise such as considerable brightness unevenness.
Therefore, it is necessary to smooth this to reduce noise during observation. The matrix operator for this smoothing process is
As shown in Figure 6, it has a 3x3 matrix configuration, and the density of the surrounding pixels is set to 1/8 by a factor of 1/8 for the center pixel of interest.
The value obtained by multiplying and adding these values is taken in as the density value of the point of interest, and this is arithmetic processed while scanning the pixels of the entire screen to smooth out variations in isolated brightness (density) between pixels. The concentration threshold value determination unit 42 determines the concentration threshold value M by subtracting an empirical value given as necessary from the road surface average concentration measurement value performed in advance. An example of the histogram of the input image and the density threshold M is shown in FIG. The concentration threshold M is
It is used in the density conversion process in the half-threshold processing unit 43 to make the pixels of the brightness portion corresponding to the road background uniform in terms of density.

The modal method, discriminant analysis method, P-tile method, differential histogram method, etc. are known as methods for determining the binarization threshold for graphic analysis in image processing technology. Therefore, the actual concentration average of a certain area of the road surface to be measured is measured in advance, and the concentration threshold determining unit 42 subtracts the empirical value set externally from the measured value of the road surface average concentration, as shown in FIG. 8(a). A density threshold M is determined on the input image histogram as follows. This half-threshold processing using the density threshold M can be said to limit the absolute density range of the cracked image, and the density data for each pixel of the smoothed input image is compared with the density threshold M as shown in Figure 8(b). However, density data of M or more is uniformly converted to density M to prevent variations in image brightness from affecting subsequent edge extraction.

In other words, there is a certain degree of difference in brightness (density) between the properties of the road surface itself, such as uneven color, and the properties of the cracks themselves, and even if there is uneven brightness on the road surface, it is relatively more noticeable than the cracks. High brightness. Therefore, the target road surface is measured in advance under actual measurement conditions, the average density is determined from, for example, 128 lines of one-dimensional image information, and the density threshold value M is determined by applying correction to this as necessary. As a result, in the half-threshold processing, for example, a density higher than the threshold M is assumed to be derived from the road surface and is converted to a uniform density, thereby eliminating noise in a pixel portion having a density higher than the threshold M. This concentration threshold value M can be modified as appropriate by adjusting the above-mentioned correction amount based on the empirical value from past measurement data.

The image information thus density-converted by the half-threshold processing unit 43 is then subjected to binarization noise processing to detect crack candidates, but the crack part has lower luminance than the surrounding road surface part. However, this is a relative brightness change, and simply performing binarization processing will not provide a satisfactory result for crack detection. Therefore, here, the edges of the cracks are detected by the edge extraction processing section 44 and then the binarization processing is performed. Common methods for extracting the edges of figures in image information include the first or second differential method, Sobel method, Roberts method, Prewitt method, and Laplacian method. In edge extraction by the unit 44, in order to perform real-time processing while taking advantage of these effects, a simpler operator in terms of hardware is defined as follows to perform edge detection processing.

That is, when the data at the point of interest is f(x, y) and the converted data is g(x, y), g(x, y)=Max{h0(x, y), h1(x ,y
), h2 (x, y), h3 (x, y)} h0 (x, y
)=|f(x,y-1)-f(x,y+1)|h1(
x,y)=|f(x+1,y-1)-f(x-1,y+
1) | h2(x,y)=|f(x-1,y)-f(x
+1,y) | h3(x,y)=|f(x-1,y-1
) - f (x+1, y+1) | This operator takes the absolute value of the difference in density between pixels adjacent to the point of interest in a total of four directions (vertical, horizontal, and diagonal) centered on the point of interest, and focuses on the maximum value among them. This is the process of converting it into point data.

The image data obtained in this way is difference value image data with the luminance difference as a scale, and if a crack exists, the value at the boundary becomes high. This difference value data is
The noise removal binarization processing unit 45 further focuses on the presence or absence of edge features and performs noise processing and binarization using a matrix operator as shown in FIG. This is an edge enhancement method that uses the center pixel of a 5 x 5 pixel matrix as the pixel of interest, and uses the value obtained by adding the densities of the pixels in eight directions around it with an interval of one pixel as the point of interest data. This enables real-time processing using hardware. Note that in FIG. 9, the operator calculates an absolute value. In this way, in the edge extraction by the edge extraction processing unit 44, the calculation using the operator shown in FIG. 9 is performed for each pixel of interest, and the data of the calculation result for each pixel of interest is saved as the conversion data of that pixel of interest. . In this way, a series of binary edge image data corresponding to the degree of density change of the contour edge of the grayscale image in the two-dimensional image information is generated together with position data. The binarized edge image data obtained by converting in this manner is then passed to the area evaluation determination processing section 46.

Each line segment image of the edge data whose features have been extracted through the series of processes described above is a crack candidate, and the area evaluation determining unit 46 evaluates the area of the crack image for each image unit for each unit section. Each time, the position coordinate value is assigned to the crack image data, and the area ratio of the crack candidate is calculated based on the ratio of the total number of pixels of each label to the unit area area, and this is compared with a preset threshold. do. Based on this comparison result, image information including position coordinates and each label coordinate data for the unit section having a crack candidate area ratio exceeding the threshold value is recorded as DAT.
Output to recorder 6.

[0036]

[Effects of the Invention] As described above, according to the present invention, the image information obtained by optically imaging the road surface is subjected to image processing optimized for crack detection and area evaluation to minimize the necessary amount of crack detection. Since it is recorded as digital data, it is possible to automatically detect cracks in the road surface efficiently using certain criteria without any restrictions on the material structure of the target road surface or the measurement conditions. It is possible to automatically detect and evaluate cracks through real-time processing using operators, and record the minimum necessary data on a portable recording medium such as DAT in real time during measurement work, along with crack position information, and take it home.

[Brief explanation of the drawing]

FIG. 1 is a schematic perspective view showing the general configuration of an apparatus according to an embodiment of the present invention.

FIG. 2 is a schematic side sectional view showing an example of the arrangement of a camera and a light source lamp in the truck hood.

FIG. 3 is a schematic plan view showing an example of alternate arrangement of light source lamps.

FIG. 4 is a block diagram showing a configuration example of a signal processing system.

FIG. 5 is a flowchart showing the flow of an algorithm for image information conversion processing in the measurement processing device.

FIG. 6 is an explanatory diagram showing an example of an operator matrix for smoothing processing.

FIG. 7 is a diagram showing an example of a density histogram of an input image.

FIG. 8(a) is a density histogram diagram for explaining density threshold determination and half-threshold processing, and FIG. 8(b) is a half-threshold processing input-output characteristic diagram.

FIG. 9 is an explanatory diagram showing an example of a matrix operator for noise removal binarization processing.

[Explanation of symbols]

1: Hood, 2: Illumination lamp, 3: Imaging camera, 4: Measurement processing device, 5: Wheels, 6: DAT recorder, 7: Traction rod, 8: Camera unit, 9: Measurement unit, 10: Power supply cart, 41 : smoothing processing unit, 42: density threshold determining unit,
43: half-threshold processing section, 44: edge extraction processing section, 45:
Noise removal binarization processing unit, 46: Area evaluation determination unit.

Claims (1)

[Claims] 1. An imaging device that outputs one-dimensional image information of a road surface in a direction intersecting the direction of movement, with a predetermined number of pixels, on a trolley that can move on a road surface, with the road surface immediately below as a field of view; A lighting device comprising: a light source disposed above a ground plane of the trolley so as to illuminate the road surface within the field of view at a low irradiation angle within a hood that shields the road surface within the field of view from the surroundings; A measurement processing device that captures the one-dimensional image information from the imaging device every predetermined travel distance and performs real-time two-dimensional image processing together with the position information of the trolley, and information output from the measurement processing device can be attached and detached. the one-dimensional image information for a predetermined number of scans in order to measure the average brightness of the road surface being measured; density threshold determining means for determining a density threshold from the average density value of image pixels for each image; means for performing half-threshold processing to convert and fix the value to the same value as the threshold;
In the two-dimensional image processing, the maximum value of the absolute value of the difference in density between pixels adjacent to the point of interest in a total of four directions (vertical, horizontal, and diagonal) centered on the point of interest with respect to the pixel of interest in scanning in the two-dimensional image processing is determined as the point of interest. edge extraction processing means that performs conversion processing into data; and for each pixel of the image data after the edge extraction processing, surrounding pixels located a predetermined pixel distance from the point of interest in a total of four directions (vertical, horizontal, and diagonal) centered on the point of interest; a noise removal processing means that performs a conversion process using the sum of the densities of the noise removal process as point of interest data; and a noise removal processing means that performs a conversion process using the sum of the densities of Area evaluation determination processing that evaluates the crack candidate area from the number of pixels of the candidate and outputs binary image information of only the unit section having a crack candidate area exceeding a predetermined threshold value to the recording device together with the position information of the unit section. A road surface crack measuring device characterized by comprising: means.