TWI579637B - Image enhancement method for lens stain detection - Google Patents

Description

Image enhancement method for lens dirty detection

The invention relates to an image enhancement method; in particular, to an image enhancement method for lens contamination detection.

During the assembly process of the lens module, due to the influence of the human component assembly ring, particles and dust in the air are likely to be attached to the optical component with the assembly action, and the particles will enter the lens as the optical component is assembled. Inside the module, the light in the lens module is blocked, which causes black spots and the like to appear in the image after the lens module is imaged.

At present, the method for detecting whether there is a particle in the lens module is that an image sensor (for example, CCD) is disposed at the imaging portion of the lens module to receive the imaging pattern of the lens module, and then the operator is borrowing When the image pattern is observed to have a particle image, the optical element in the lens module is determined to be contaminated by the particles.

However, if the imaging pattern is directly observed, the particle size is too small or the particle image is too close to the color of other images, so that the operator can easily ignore the dirty image in the imaging pattern, and cannot find the defective lens module.

In view of the above, it is an object of the present invention to provide an image enhancement method for lens contamination detection.

In order to achieve the above object, the lens provided by the present invention is dirty. The image enhancement method for detecting, wherein the image is taken from a color image imaged by the lens during the stain detection process, and the color image presents an optical lens in the lens and a dirty surface on the optical lens The image enhancement method includes the following steps: A. extracting a red layer and a green layer in the color image, and converting the red layer and the green layer into a first grayscale image and a second grayscale image, respectively. . B. multiplying the grayscale values of each pixel in the first grayscale image and the second grayscale image by a preset weight at least once to improve the first grayscale image and the second grayscale Grayscale contrast of the order image. C. The first gray scale image after the gray scale contrast is improved, and the second gray scale image after the gray scale contrast is superimposed to obtain a superimposed image. D. Improve the grayscale contrast in the superimposed image. E. Binding the superimposed grayscale image to obtain a black and white image, and the optical lens and the stain present in the black and white image exhibit opposite colors.

The effect of the present invention is that it is possible to quickly determine whether or not there is any stain in the color image. In addition, if the operator is required to judge whether the black and white image is dirty, it can also quickly and accurately distinguish the dirt due to the high contrast of the image to be tested.

12‧‧‧Dirty

14‧‧‧Optical lenses

16‧‧‧ Around the optical lens

S100, S200, S300‧‧‧ steps

S202~S220‧‧‧Steps

S400, S402‧‧‧ steps

1 is a flow chart of an image enhancement method for lens contamination detection according to a preferred embodiment of the present invention.

Figure 2 is a color photograph.

Figure 3 shows the image to be tested.

4 is a flow chart of the first, third, fifth, and seventh algorithms of step S200 of FIG. 1 described above.

Figure 5 is a superimposed image.

Figure 6 shows the image of the edge-finding algorithm.

FIG. 7 is an image after the step S210 is performed.

FIG. 8 is an image after the step S214 is performed.

FIG. 9 is a flow chart of a method for manufacturing an edge vignetting image.

Figure 10 is an image of the edge vignetting.

FIG. 11 is a flow chart of the second algorithm of step S200 in FIG. 1 described above.

FIG. 12 is a flow chart of the fourth algorithm of step S200 in FIG. 1 described above.

FIG. 13 is a flow chart of the sixth algorithm of step S200 in FIG. 1 described above.

FIG. 14 is a flow chart of the eighth algorithm of step S200 in FIG. 1 described above.

In order to explain the present invention more clearly, the preferred embodiment is described in detail with reference to the accompanying drawings. FIG. 1 is a flow chart of an image enhancement method for lens contamination detection according to a preferred embodiment of the present invention.

Step S100: placing an image sensor on one side of the lens module to capture the optical lens in the lens, and causing the image sensor to obtain a color image according to the image of the optical lens (see FIG. 2). As shown in the figure), the color image can be image processed later using a general computer. In addition to presenting the optical lens 14 within the lens, the color image also has a stain 12 on the optical lens. The image sensor is a CCD. In other embodiments, it may also be an image sensor made by CMOS manufacturing technology, which is not limited thereto.

Step S200: Perform image processing on the color image to obtain a to-be-tested image (as shown in FIG. 3), and a part of the image to be tested is a black-colored tile, and the other part is white. The tile has a plurality of white tiles and is dispersed in the image to be tested.

Step S300: analyzing the size of each white tile in the image to be tested. If the size of a white tile is within the set interval, it is determined that the white tile is dirty.

In the embodiment of the present invention, the size, the length, and the width of the pixel (pixel) of the white tile are analyzed in an interval to determine whether the white tile is For dirty, for example, set the area between 30 and 1100, the length to 5 to 635, and the width to 5 to 475. In other implementations, the setting of parameters such as the shape or position of the white tile may be further determined to determine whether the white tile is dirty.

The above is the image detecting method of the present invention, wherein in step 200, different algorithms can be used for different positions of dirt and dirt and contrast exhibited by the optical lens, and eight different algorithms will be described below: As shown in FIG. 4, it is a flowchart of the first algorithm in step 200.

Step S202: Extract a red layer and a green layer in the color image, and convert the red layer and the green layer into a first grayscale image and a second grayscale image. The reason why the red layer and the green layer are extracted is because the color difference between red and green is large, so if there is dirt in the lens module, it is easy to be presented in the red layer or the green layer.

Step S204: Multiply the grayscale value of each pixel in the first grayscale image and the second grayscale image of the green layer by a preset weight at least once to improve the first grayscale image and the second Grayscale contrast of grayscale images. In this embodiment, the grayscale value of each pixel in the red layer and the green layer is multiplied by the corresponding weight three times, and the weight is a 3×3 digital filter, as follows:

Step S206: superimposing the first grayscale image after the grayscale contrast and the second grayscale image of the grayscale contrast to obtain a superimposed image (as shown in FIG. 5), and the superimposing The grayscale contrast of the optical lens and the dirty image presented in the image is higher than that of the optical lens and the dirty image presented in the color image. In addition, the optical lens presented in the superimposed image and the image around the optical lens are also highly contrasted.

Step S208: Perform a Sobel operator on the superimposed image to find the optical lens and the dirty edge presented in the superimposed image, and present in white (as shown in FIG. 6). .

Step S210: Multiply the grayscale value of each pixel in the superimposed image after performing the edge-finding algorithm by a preset weight of 10 times, and improve the grayscale contrast (as shown in FIG. 7). In addition, the weights described in this step are the same as the weights of step S204, and are not described again.

Step S212: Binarize the superimposed grayscale image after step 210 to obtain a black and white image, and the optical lens and the dirtyness presented in the black and white image exhibit opposite colors. In this embodiment, the color of the stain presented in the black and white image is mostly white, and the color of the optical lens presented in the black and white image is mostly black. The threshold value of the binarized grayscale value is set to 134 in this embodiment, but other embodiments may be adjusted according to the average grayscale value of the superimposed image, and not limited thereto.

Step S214: Perform a close operation on the black and white image to filter out noise and fill holes in the dirty image presented in the black and white image, so that all pixels in the dirty image are white. (As shown in Figure 8). In the present embodiment, the closing algorithm is performed 15 times.

Step 216: The black and white image is superimposed with an edge vignetting image, and the superimposed image is closed for 5 times to obtain an image to be tested (as shown in FIG. 3). The edge vignetting image in the above refers to a kind of light The lens and the image surrounding the periphery of the optical lens 16 (shown in Figure 2) are clearly distinguishable images. In this step, the black and white image is superimposed with an edge vignetting image to ensure that the image present in the image to be tested is dirty and the color around the optical lens is white, and the color of the optical lens is black. In order to facilitate the analysis of the dirty.

As shown in FIG. 9 , a flow chart of a method for manufacturing an edge vignetting image includes the following steps:

Step 400: Extract the red layer in the color image, and convert the red layer into the first grayscale image.

Step 402: Binarize the first grayscale image, and then perform reverse processing to make the optical lens appear black, and the periphery of the optical lens is white to obtain the edge vignetting image (as shown in FIG. 10).

As shown in FIG. 11, it is a flowchart of the second algorithm in step 200.

The difference between the first algorithm and the second algorithm is that in step S210, the preset weight is multiplied by 15 times, and in step S212, the threshold value of the binarized gray level value is set to 223, and The closing algorithm is performed 17 times in step S214. Further, step S218 is further added between step S210 and step S212.

Step S218: performing a dilation algorithm (Dilate) 30 times on the superimposed image, and then performing an erosion algorithm (Erode) 30 times to improve the gray scale contrast.

As shown in Figure 4, the third algorithm is explained.

The third algorithm is similar to the first algorithm in that it is multiplied by the corresponding weight 15 times in step S210, and the binarized gray level value threshold in step S212 is set to 255 and The closing algorithm is performed 17 times in step S214.

As shown in FIG. 12, it is the fourth algorithm in step 200. flow chart.

The fourth algorithm is similar to the first algorithm, and the difference is that step S210 is not performed, and in step S210, each step is multiplied by a preset weight of 15 times, and the gray scale value valve is binarized in step S212. The value is set to 255 and the closing algorithm is performed 5 times in step S214. Further, step S220 is further added between step S206 and step S210.

Step S220: Multiply the grayscale value of each pixel in the superimposed image by a preset weight once. The weight in this step is a 3 × 3 digital filter:

As shown in Figure 4, the fifth algorithm is explained.

The fifth algorithm is similar to the first algorithm in that, in step S204, each of the preset weights is multiplied twice and the closing algorithm is performed 10 times in step S214.

As shown in FIG. 13, it is a flowchart of the sixth algorithm in step 200.

The sixth algorithm is similar to the second algorithm, the difference is that step S208 is not performed, and after step S206 is completed, step S210 is directly performed, wherein in step S210, each of the preset weights is multiplied. 10 times. Further, step S214 is not performed, and after step S212 is completed, step S216 is directly performed, in which the threshold value of the grayscale value binarized in step S212 is set to 223.

As shown in Figure 4, the seventh algorithm is illustrated.

The seventh algorithm is similar to the first algorithm, and the difference is that in step S210, each of the preset weights is multiplied five times, and the threshold value of the grayscale value binarized in step S212 is set to 88 and The closing algorithm is performed 5 times in step S214.

As shown in FIG. 14, it is a flowchart of the eighth algorithm in step 200.

The eighth algorithm is similar to the first algorithm, with the difference that the weights multiplied in step S204 are:

And the grayscale value of each pixel in the red layer and the green layer is multiplied by the preset weight once.

Further, the eighth algorithm does not perform step S206 and step S214 in the first algorithm, and the binarized grayscale value threshold is set to 255 in step S212.

The first to fourth algorithms described above mainly detect color images having images with high contrast and dirty lenses, but are subdivided into four different algorithms depending on the degree of contrast. The order is the first to fourth algorithms. The fifth and sixth algorithms mainly detect color images with relatively low contrast and optical lens images, but subdivide two different algorithms depending on the degree. The seventh and eighth algorithms mainly detect the color image of the dirty spot on the edge of the optical lens, and subdivide into two different algorithms according to the distance between the dirty and the edge of the optical lens.

In addition, in addition to the above steps, the image enhancement method for the lens contamination detection of the present invention may first determine whether the average grayscale value of the red layer is less than 128 to ensure that the subsequent steps can determine whether there is any contamination.

In summary, the image detecting method of the present invention can quickly determine whether there is any stain in the color image to find a defective lens module. In addition, if the operator needs to judge whether the white block in the image is dirty, it can also be fast and accurate due to the high contrast of the image to be tested (ie, the color difference between the black block and the white block). Distinguish the dirty position.

The above is only a preferred embodiment of the present invention, and equivalent changes to the scope of the present invention and the scope of the patent application are intended to be included in the scope of the present invention.

S202~S216‧‧‧Steps

Claims (7)

An image enhancement method for lens contamination detection, wherein the image is taken from a color image imaged by the lens during the stain detection process, and the color image presents an optical lens in the lens and is located on the optical lens The image enhancement method comprises the following steps: A. extracting a red layer and a green layer in the color image, and converting the red layer and the green layer into a first grayscale image and a first a gray scale image; B. multiplying the gray scale values of each pixel in the first gray scale image and the second gray scale image by a preset weight at least once to improve the first gray scale image Corresponding to the gray scale of the second grayscale image; C. superimposing the grayscale contrasted first grayscale image and increasing the grayscale contrasted second grayscale image overlay to obtain a superimposed image; D. improving the grayscale contrast in the superimposed image; and E. binarizing the superimposed grayscale contrast image to obtain a black and white image, and the optical lens presented in the black and white image and This stain shows the opposite color. The image enhancement method of claim 1, wherein the black and white image further comprises an edge tile region, the image detection method further comprising the steps of: binarizing the first grayscale image and performing image inversion, and Obtaining an edge vignetting image; and superimposing the black and white image and the edge vignetting image, and performing a closing algorithm to obtain a to-be-tested image, and the optical lens in the image to be tested and surrounding the optical lens The image shows the opposite color. The image enhancement method according to claim 2, further comprising the step of: analyzing the size of the image presented by the dirt in the image to be tested to determine that the image is dirty. The image enhancement method of claim 1, wherein the step D further comprises performing a edge finding algorithm on the superimposed image to find an edge of the optical lens and the dirty edge presented in the superimposed image. Then, the grayscale value of each pixel in the superimposed image after performing the edge-finding algorithm is multiplied by a preset weight to improve the grayscale contrast. The image enhancement method of claim 1, wherein after the step D, the superimposed algorithm and the erosion algorithm are sequentially performed on the superimposed image to improve the contrast of the superimposed image. The image enhancement method of claim 1, wherein the step D further comprises multiplying the grayscale values of each pixel in the superimposed image by a preset weight to improve the contrast of the superimposed image. The image enhancement method of claim 1, wherein the black and white image is closed after the step E to filter out the noise in the image that is dirty.