CN111091556B

CN111091556B - Automatic detection method for scratches on surface of automobile instrument panel

Info

Publication number: CN111091556B
Application number: CN201911285288.0A
Authority: CN
Inventors: 童莹; 赵曼雪; 曹雪虹
Original assignee: Nanjing Institute of Technology
Current assignee: Nanjing Institute of Technology
Priority date: 2019-12-13
Filing date: 2019-12-13
Publication date: 2022-04-12
Anticipated expiration: 2039-12-13
Also published as: CN111091556A

Abstract

An automatic detection method for scratches on the surface of an automobile instrument panel relates to the technical field of automatic image detection methods. The invention comprises the following steps: 1) inputting an instrument panel image, preprocessing the instrument panel image and removing noise interference; 2) acquiring a complex printing icon/symbol template at the center of an instrument panel image; 3) matching a printing icon/symbol template; 4) acquiring an area in a circle of the instrument panel by adopting a contour detection method; 5) acquiring a connected region of a region in an instrument panel circle by adopting a contour detection method; 6) judging whether the connected domain of the printed icon has a scratch or not; 7) and identifying the detected complicated printed icon/symbol scratch at the center, background scratch and simple printed icon/symbol scratch at the edge in the image. The scratch detection device has the advantages that the scratch of the automobile instrument panel is automatically detected by the computer, the defect of traditional manual detection can be effectively overcome, the printing quality of the instrument panel is guaranteed, the working efficiency is improved, and the production cost is reduced.

Description

Automatic detection method for scratches on surface of automobile instrument panel

Technical Field

The invention relates to the technical field of automatic detection methods for scratches on the surface of an automobile instrument panel.

Background

In recent years, the automobile industry is rapidly developed, the automobile sales volume in China exceeds 2800 thousands in 2018, and the automobile industry is the first in the world in 10 years. As an indispensable guarantee for safe driving of automobiles, the instrument panel is a very important automobile part, and the correctness and the clarity of printing of each warning symbol and icon on the panel are important prerequisites for ensuring the quality and the safety of automobile products.

Scratches on the dashboard of the car may be present in the target area or background area where the scratches may be present on either a complex printed icon/symbol in the center or a simple printed icon/symbol at the edge. At present, the automobile industry in China generally adopts a manual mode to detect the printing quality of instrument panels, the method is greatly influenced by subjective factors of people and external environment, when large batches of instrument panels are detected in the printing quality, the long-time and repeated work easily causes visual fatigue of inspectors, causes false detection and missed detection, and has the defect of low detection efficiency.

Disclosure of Invention

The invention aims to provide an automatic detection method for scratches on the surface of an automobile instrument panel, which adopts a computer to automatically detect scratches of the automobile instrument panel, can effectively overcome the defects of the traditional manual detection, not only ensures the printing quality of the instrument panel, but also improves the working efficiency and reduces the production cost.

An automatic detection method for scratches on the surface of an automobile instrument panel is characterized by comprising the following steps:

1) inputting an instrument panel image, preprocessing the instrument panel image and removing noise interference;

2) acquiring a complex printing icon/symbol template at the center of an instrument panel image;

3) adopting a template matching method to carry out template matching on the printed icons/symbols, judging whether scratches exist in the matched complex printed icon/symbol area at the center, if so, marking out the complex printed icon/symbol area, otherwise, removing the matched complex printed icon/symbol at the center from the instrument panel image;

the template matching method is specifically realized as follows:

and sliding the template image on the original image, and moving one pixel position at a time according to the sequence from left to right and from top to bottom. At each position, the matching metric value R (x, y) is calculated using the following formula, where T (x ', y') is the gray-level value of the template image at (x ', y'), and I (x + x ', y + y') is the gray-level value of the original image at (x + x ', y + y').

And counting the sum of squares of the gray difference values of all pixels in the template image and the pixels of the corresponding original image to obtain a matching metric value R (x, y), wherein the smaller the matching value is, the more similar the area corresponding to the original image and the template image are, and otherwise, the more dissimilar the area corresponding to the original image and the template image are.

4) Acquiring an area in a circle of the instrument panel by adopting a contour detection method;

extracting the outlines of all connected domains in the instrument panel image by an outline detection method, then calculating the circumferences of all the outlines, sorting the outlines according to the descending order of the circumferences, and selecting the outline corresponding to the second large circumference value, namely the outline of the circular area in the instrument panel in the graph. Setting the gray value of the pixel outside the outline as 0, and obtaining the image of the area inside the instrument panel circle, as shown in fig. 7.

5) Acquiring a connected region of a region in an instrument panel circle by adopting a contour detection method; judging whether the connected domain is the connected domain of the printed icon at the edge, if so, judging whether the connected domain of the printed icon has a scratch, and if not, judging that the connected domain of the printed icon has the background scratch;

and extracting the outlines of all target areas in the instrument panel circle by adopting an outline detection method, wherein the outlines are represented by a series of pixel point sets, and one pixel point set corresponds to one outline. The region outlined by these pixel point sets is the connected domain. Calculating the average gray value of all pixels in each connected domain, and if the average gray value is between 75 and 160, judging that the region is a background scratch at a non-target region; otherwise, the printing icon connected domain.

6) Judging whether the connected domain of the printed icons has the scratch or not, if so, displaying the simple printed icons/symbols scratch at the edge;

and further judging whether scratches exist in the residual communication area at the edge, carrying out size normalization on the residual communication area image, carrying out binarization operation on the normalized image, counting the number of target pixels in each line in each binary image, comparing the number of the target pixels with the number of the target pixels of the binary image of the corresponding edge standard simple icon/symbol template line by line, and if the difference of the number of the target pixels of n continuous lines is greater than a threshold value T, wherein n is greater than or equal to 5, and T is greater than or equal to 6, judging that the scratches exist in the small communication area.

7) Identifying in the image the complex printed icon/symbol scratch at the center detected in step 3), the background scratch detected in step 5), and the simple printed icon/symbol scratch at the edge detected in step 6).

By adopting the technical scheme, compared with the prior art, the invention has the following advantages:

the invention detects the printed icons/symbols in the input instrument dial circle, and marks the scratches in the target area (the central complex printed icon/symbol), the non-target area (the background) and the target area (the edge simple printed icon/symbol) by using a template matching corner point detection method and a contour detection method. The whole process is automatically completed through a computer, so that the detection precision is improved, the detection time is shortened, and the printing quality detection automation of the automobile instrument panel is realized. False detection and missing detection caused by human eye detection are effectively avoided, the working efficiency is improved, and the production cost is reduced.

Drawings

FIG. 1 is an overall flow chart of the present invention.

Fig. 2a is a schematic illustration of a scratch at the target area (central complex printed icon/symbol).

Fig. 2b is a schematic illustration of a scratch at a non-target area (background).

Figure 2c is a schematic illustration of the scratch at the target area (simply printed icon/symbol at the edge).

Fig. 3 is a schematic illustration of a complex printed icon/symbol template.

Figure 4 is a schematic view of a simple edge printed icon/symbol template.

Fig. 5a is a schematic diagram of the matching of the scratched icon/symbol image.

Fig. 5b is a schematic diagram of a standard complex printed icon/symbol template corresponding to fig. 5 a.

Fig. 6a is a schematic diagram of fig. 2a after the image is removed.

Fig. 6b is a schematic diagram of fig. 2b after the image is removed.

Fig. 6c is a schematic diagram of fig. 2c after the image is removed.

Fig. 7a is a schematic view of the circled area in fig. 2 a.

Fig. 7b is a schematic view of the circled area in fig. 2 b.

Fig. 7c is a schematic view of the circled area in fig. 2 c.

Fig. 8a is a scratch at the non-target area (background) detected in fig. 7 b.

Fig. 8b is a scratch at the non-target area (background) detected in fig. 7 c.

Fig. 9a is a schematic illustration showing the scratched area in fig. 7 c.

Fig. 9b is a schematic diagram of a standard simple printed icon/symbol template corresponding to fig. 9 a.

Fig. 10a is a schematic illustration of a scratch at the target area (central complex printed icon/symbol).

Fig. 10b is a schematic illustration of a scratch at a non-target area (background).

Fig. 10c is a schematic illustration of the scratch at the target area (simply print icon/symbol at the edge).

Detailed Description

The technical scheme of the invention is explained in the following with the accompanying drawings:

as shown in fig. 1, an automatic detection method for scratches on the surface of an automobile instrument panel includes the following steps:

3) performing template matching on the printed icons/symbols by adopting a template matching method, and removing the matched complex printed icons/symbols at the center from the instrument panel image;

the template matching method is specifically realized as follows:

the template image of fig. 3 is slid on the original image of fig. 2, and the positions are shifted one pixel at a time in the order of left to right and top to bottom. At each position, the matching metric value R (x, y) is calculated using the formula, where T (x ', y') is the gray-level value of the template image at (x ', y'), and I (x + x ', y + y') is the gray-level value of the original image at (x + x ', y + y').

the specific method comprises the following steps:

6) Judging whether the connected domain of the printed icons has a scratch result, if so, displaying simple printed icons/symbols scratches at the edge;

the specific method comprises the following steps:

In the step 1), the input instrument panel image is manually cut, and a plurality of small icons which belong to the same printed icon/symbol but are not communicated in area are integrally cut to manufacture a template; meanwhile, a contour detection method is adopted to obtain simple icon/symbol templates with communicated edge areas.

Acquiring a complex printed icon/symbol template at the center of an instrument panel image in step 2) of the invention, carrying out template matching on an automobile instrument panel image, and carrying out size normalization on the matched icon/symbol image; and (3) carrying out binarization operation on the normalized image, counting the number of target pixels of each line in each binary image, comparing the counted number of the target pixels of each line with the number of the target pixels of each line of the corresponding standard template binary image, and if the difference of the number of the target pixels of n continuous lines is greater than a threshold value T, wherein n is greater than or equal to 5, and T is greater than or equal to 6, judging that the matched printed icon/symbol has a scratch.

In the step 5), all connected regions in the graph are obtained by adopting a contour detection method, the contour of each target region in the instrument panel circle is extracted by adopting a contour detection method, the contours are represented by a series of pixel point sets, and one pixel point set corresponds to one contour. The region outlined by these pixel point sets is the connected domain. Calculating the average gray value of each connected region, and if the average gray value is between 75 and 160, judging that the region is a scratch at a non-target region; and further judging whether scratches exist in the residual communication area at the edge, carrying out size normalization on the residual communication area image, carrying out binarization operation on the normalized image, counting the number of target pixels in each line in each binary image, comparing the number of the target pixels with the number of the target pixels of the binary image of the corresponding edge standard simple icon/symbol template line by line, and if the difference of the number of the target pixels of n continuous lines is greater than a threshold value T, wherein n is greater than or equal to 5, and T is greater than or equal to 6, judging that the scratches exist in the small communication area.

As shown in fig. 2a, 2b, and 2c, three types of car dashboard images are inputted, wherein fig. 2a shows that the scratch is at the target area (the central complex printed icon/symbol), fig. 2b shows that the scratch is at the non-target area (background), and fig. 2c shows that the scratch is at the target area (the edge simple printed icon/symbol). Since there are many kinds of printed icons/symbols in the central area of the dashboard, and the same printed icon/symbol is composed of a plurality of small icons, it is not convenient to judge the whole connected domain, therefore, a manual cutting method is adopted here to cut the whole of the plurality of small icons which belong to the same printed icon/symbol but are not connected in the area, and make into a template, as shown in fig. 3. Meanwhile, a simple icon/symbol template with connected edge regions is obtained by adopting a contour detection method, as shown in fig. 4.

And (3) performing template matching on the automobile instrument panel image by using the complex printed icon/symbol template (as shown in figure 3) at the center acquired in the step (2), and performing size normalization on the matched icon/symbol image. And (3) carrying out binarization operation on the normalized image, counting the number of target pixels in each line of each binary image, comparing the counted number of the target pixels in each line with the number of the target pixels in each line of the corresponding standard template binary image, and if the difference between the number of the target pixels in n continuous lines (n is more than or equal to 5) is more than a threshold value T (T is more than or equal to 6), judging that the matched printing icon/symbol has a scratch, as shown in fig. 5a and 5 b. Fig. 5a is a matching image of a scratched icon/symbol. Fig. 5b is a schematic diagram of a standard complex printed icon/symbol template corresponding to fig. 5 a.

Removing the complex icon/symbol at the center obtained by matching in step 3 from the dashboard image, as shown in fig. 6a, 6b, and 6c, where fig. 6a is a display diagram of fig. 2a after the image is removed, fig. 6b is a display diagram of fig. 2b after the image is removed, and fig. 6c is a display diagram of fig. 2c after the image is removed. And obtaining the area in the instrument panel circle by adopting a contour detection method, extracting the contour of each connected area in the instrument panel image by adopting the contour detection method, then calculating the perimeter of each contour, sequencing the contours in a descending order according to the size of the perimeter, and selecting the contour corresponding to the second large perimeter value, namely the contour of the area in the instrument panel circle in the image. Setting the gray value of the pixel outside the outline to 0, to obtain the image of the area inside the circle of the instrument panel, as shown in fig. 7a, 7b, and 7c, wherein fig. 7a is a schematic display diagram of the area inside the circle of fig. 2a, fig. 7b is a schematic display diagram of the area inside the circle of fig. 2b, and fig. 7c is a schematic display diagram of the area inside the circle of fig. 2 c.

On the basis of fig. 7, all connected small regions in the map are obtained by using a contour detection method, an average gray value of each connected small region is calculated, and if the average gray value is between 75 and 160, the region is determined to be a scratch at a non-target region (background), as shown in fig. 8a and 8 b. Where fig. 8a is the scratch at the non-target area (background) detected in fig. 7b, and fig. 8b is the scratch at the non-target area (background) detected in fig. 7 c.

And further judging whether scratches exist in the residual connected small areas at the edges. Performing size normalization on the remaining connected small region images, performing binarization operation on the normalized images, counting the number of target pixels in each line of each binary image, and comparing the number of target pixels with the number of target pixels of a corresponding edge standard simple icon/symbol template (such as fig. 4) line by line, if the difference between the number of target pixels of n consecutive lines (n is greater than or equal to 5) is greater than a threshold value T (T is greater than or equal to 6), determining that a scratch exists in the connected small region, as shown in fig. 9a and 9b, wherein fig. 9a is the scratch detected in fig. 7c at the target region (edge simple printed icon/symbol), and fig. 9b is the standard simple printed icon/symbol template corresponding to fig. 9 a.

Finally, the scratches detected in steps 3, 5 and 6 are marked with red frames on the original drawing, as shown in fig. 10, where fig. 10a is a graph of the detection result of the scratch in the target area (the center complex printed icon/symbol), fig. 10b is a graph of the detection result of the scratch in the non-target area (the background), and fig. 10c is a graph of the detection result of the scratch in the target area (the edge simple printed icon/symbol). Note that fig. 10c also includes the scratch detection result at the non-target region (background).

The invention detects the printed icons/symbols in the input instrument dial circle, and marks the scratches in the target area (the central complex printed icon/symbol), the non-target area (background) and the target area (the edge simple printed icon/symbol) by using a template matching method and a contour detection method. The whole process is automatically completed through a computer, so that the detection precision is improved, the detection time is shortened, and the printing quality detection automation of the automobile instrument panel is realized.

Claims

1. An automatic detection method for scratches on the surface of an automobile instrument panel is characterized by comprising the following steps:

3) adopting a template matching method to carry out template matching on the printed icons/symbols, judging whether scratches exist in the matched complex printed icon/symbol area at the center, if so, marking out the complex printed icons/symbols, and if not, removing the matched complex printed icons/symbols at the center from the instrument panel image;

the template matching method is specifically realized as follows:

sliding the template image on the original image, moving one pixel position at a time according to the sequence from left to right and from top to bottom, and calculating a matching metric value R (x, y) at each position by using the following formula, wherein T (x ', y') is the gray value of the template image on (x ', y'), and I (x + x ', y + y') is the gray value of the original image on (x + x ', y + y');

counting the sum of squares of gray difference values of all pixels in the template image and pixels of the corresponding original image to obtain a matching metric value R (x, y), wherein the smaller the matching value is, the more similar the area corresponding to the original image and the template image are, and otherwise, the more dissimilar the area corresponding to the original image and the template image are;

extracting the outlines of all connected areas in the instrument panel image through an outline detection method, then calculating the perimeters of all the outlines, sorting the perimeters of the outlines according to descending order of the sizes, and selecting the outline corresponding to the second large perimeter value, namely the outline of the area in the instrument panel circle in the graph; setting the gray value of the pixel outside the outline as 0 to obtain an area image inside the instrument panel circle;

extracting the outlines of all target areas in the instrument panel circle by adopting an outline detection method, wherein the outlines are represented by a series of pixel point sets, one pixel point set corresponds to one outline, the areas outlined by the pixel point sets are connected areas, the average gray value of all pixels in each connected area is calculated, and if the average gray value is 75-160, the area is judged to be a background scratch in a non-target area; otherwise, the printing icon is a connected domain;

6) judging whether the connected domain of the printed icon has a scratch or not; if so, then a simple print icon/symbol scratch at the edge is displayed;

judging whether scratches exist in the remaining connected regions at the edges, carrying out size normalization on the images of the remaining connected regions, carrying out binarization operation on the normalized images, counting the number of target pixels in each line of each binary image, comparing the number of the target pixels with the number of target pixels of the binary images of the corresponding standard simple icon/symbol templates at the edges line by line, and if the difference of the number of the target pixels of n continuous lines is greater than a threshold value T, wherein n is greater than or equal to 5, and T is greater than or equal to 6, judging that the scratches exist in the small connected regions;

2. The method according to claim 1, wherein in step 1), the inputted dashboard image is manually cut, and a plurality of small icons belonging to the same printed icon/symbol but not connected in area are integrally cut to form a template; meanwhile, a contour detection method is adopted to obtain simple icon/symbol templates with communicated edge areas.

3. The method for automatically detecting scratches on the surface of an automobile instrument panel according to claim 1, wherein the complex printed icon/symbol template obtained in the center of the instrument panel image in step 2) is used for template matching of the automobile instrument panel image, and the size normalization of the matched icon/symbol image is performed; and (3) carrying out binarization operation on the normalized image, counting the number of target pixels of each line in each binary image, comparing the counted number of the target pixels of each line with the number of the target pixels of each line of the corresponding standard template binary image, and if the difference of the number of the target pixels of n continuous lines is greater than a threshold value T, wherein n is greater than or equal to 5, and T is greater than or equal to 6, judging that the matched printed icon/symbol has a scratch.

4. The method according to claim 1, wherein in the step 5), a contour detection method is used to obtain connected regions of the area in the dashboard circle, an average gray value of each connected region is calculated, and if the average gray value is between 75 and 160, the area is determined to be the scratch in the non-target area; and further judging whether scratches exist in the residual communication area at the edge, carrying out size normalization on the residual communication area image, carrying out binarization operation on the normalized image, counting the number of target pixels in each line in each binary image, comparing the number of the target pixels with the number of the target pixels of the binary image of the corresponding edge standard simple icon/symbol template line by line, and if the difference of the number of the target pixels of n continuous lines is greater than a threshold value T, wherein n is greater than or equal to 5, and T is greater than or equal to 6, judging that the scratches exist in the small communication area.