TWI738510B - Image overlay method of semiconductor element - Google Patents

Abstract

A method for superimposing images of a semiconductor element, which is suitable for superimposing an image of a semiconductor element, is implemented by an image superimposing device, the image superimposing device includes an image capturing unit, and an electrical connection to the image A processing unit of a photographing unit, the method includes the following steps: (A) The image photographing unit photographs the semiconductor device according to a plurality of different photographing parameter data, so as to obtain a plurality of photographs corresponding to the photographing parameter data and having the same resolution. Photographed images; and (B) the processing unit superimposes the photographed images to obtain a superimposed image. The superimposed image can reflect the difference in height and reflectance of the surface of the semiconductor element.

Description

Image overlay method of semiconductor element

The present invention relates to an image overlay method, in particular to a semiconductor element image overlay method.

With the vigorous development of semiconductor manufacturing processes, wafers are made smaller and smaller, and the accuracy of stacking between layers is becoming more and more important. Because the key size is small, the offset that can be tolerated between relative layers has also changed. Small, so the industry has more and more stringent requirements on the contour precision of various components.

In order to detect or measure the contour of the component, the existing method is to continuously take multiple images of the component and superimpose these images to produce a superimposed image with better quality, and then superimpose the superimposed image. The image is inspected or measured.

The existing images used for superimposing are usually obtained using a single setting. For example, the focal lengths used to capture the images are all the same and the exposure time used to capture the images is the same. However, the image with a single setting cannot reflect the difference in height and reflectance of the surface of the workpiece, so it is necessary to propose a solution.

Therefore, the object of the present invention is to provide a method for superimposing semiconductor device images that can reflect the difference in height and reflectance of the surface of the semiconductor device.

Therefore, the semiconductor element image superimposing method of the present invention is suitable for superimposing the image of a semiconductor element, and is implemented by an image superimposing device that includes an image capturing unit and an electrical connection The processing unit of the image capturing unit. The method includes a step (A) and a step (B).

In this step (A), the image capturing unit photographs the semiconductor device according to a plurality of different photographing parameter data to obtain a plurality of photographed images corresponding to the photographing parameter data and having the same resolution.

In this step (B), the processing unit superimposes the photographed images to obtain a superimposed image.

The effect of the present invention is that the photographed images with different characteristics are superimposed by the processing unit, so that the superimposed image obtained can reflect the difference in height and reflectance of the surface of the semiconductor element .

Before the present invention is described in detail, it should be noted that in the following description, similar elements are denoted by the same numbers.

Referring to FIG. 1, an image superimposing device 1 used to implement an embodiment of the semiconductor device image superimposing method of the present invention is illustrated. The image superimposing device 1 includes an image capturing device for capturing a semiconductor device. Unit 11, and a processing unit 12 electrically connected to the image capturing unit 11.

Referring to FIGS. 1 and 2, this embodiment of the semiconductor device image superimposing method of the present invention is suitable for superimposing the image of the semiconductor device. The steps included in this embodiment are described below.

In step 21, the image capturing unit 11 photographs the semiconductor device according to a plurality of different photographing parameter data, so as to obtain a plurality of photographed images respectively corresponding to the photographing parameter data and having the same resolution. It is worth noting that in this embodiment, each shooting parameter data includes a focal length (Focal length) parameter, an aperture (Aperture) parameter, and a shutter speed (Shutter speed) parameter, so that the captured images have Different sharpness, texture, brightness, and color, but not limited to this.

In step 22, the processing unit 12 performs image processing on the captured images to obtain multiple image-processed images corresponding to the captured images.

With reference to FIG. 3, step 22 includes sub-steps 221 to 223, and the sub-steps included in step 22 are described below.

In step 221, for each captured image, the processing unit 12 smoothes the captured image to obtain a smoothed image.

In step 222, for each captured image, the processing unit 12 subtracts the corresponding smoothed image from the captured image to obtain an edge detail image.

In step 223, for each captured image, the processing unit 12 superimposes the captured image and its corresponding edge detail image to obtain an image-processed image corresponding to the captured image.

It is worth noting that, in this embodiment, the image processing performed by the processing unit 12 is image sharpening processing, but it is not limited to this.

In step 23, the processing unit 12 grayscales and reduces the processed images to obtain multiple grayscale reduced images corresponding to the processed images and have the same resolution. picture. It is worth noting that, in this embodiment, the processing unit 12 gray-scales the processed images, and uses a neighbor correlation interpolation method to gray-scale the processed images. The image side length is reduced in proportion, but not limited to this.

， 其中，

為該灰階縮小圖像的該等像素中之一目標像素的像素位置，

為該目標像素的像素位置對應的能量梯度，

為該目標像素的鄰域像素之像素位置，

為該目標像素的像素值，

為該目標像素的鄰域像素的像素值。 In step 24, for each gray-scale reduced image, the processing unit 12 obtains a pixel position including a plurality of pixels corresponding to the gray-scale reduced image according to the multiple pixels of the gray-scale reduced image. The energy gradient group of the energy gradient. For each reduced gray scale image, the energy gradients included in the energy gradient group corresponding to the reduced gray scale image are as follows:

, in,

Reduce the pixel position of one of the pixels of the grayscale image,

Is the energy gradient corresponding to the pixel position of the target pixel,

Is the pixel position of the neighboring pixel of the target pixel,

Is the pixel value of the target pixel,

Is the pixel value of the neighboring pixel of the target pixel.

，

，

， 其中，

為該目標像素的鄰域像素之數量。 It should be noted that, in other embodiments, the energy gradients included in the energy gradient group corresponding to the grayscale reduced image can also be obtained by other calculation methods. The following will list other calculations corresponding to the grayscale reduced image The energy gradient group includes formulas for the energy gradients.

,

,

, in,

Is the number of neighboring pixels of the target pixel.

It should be noted again that the neighboring pixels of the target pixel can be 4-connected or 8-connected as the standard, where if it is 4-connected, the neighboring pixels of the target pixel are at the top, bottom, left, and right positions of the target pixel; if it is 8 If the target pixel is connected, the neighboring pixels of the target pixel are the positions of 4 pixels that also include its diagonal. In detail, if the target pixel is located at the 4 vertices of the image, the number of 4-connected neighboring pixels is 2. The 8-connected neighborhood pixels are 3; if the target pixel is located on the 4 borders of the image but is not a vertex, the number of 4-connected neighborhood pixels is 3, and the 8-connected neighborhood pixels are 5; if the target The pixel is located at any position within the four boundaries of the image, the number of its 4-connected neighboring pixels is 4, and its 8-connected neighboring pixels are 8.

In step 25, the processing unit 12 obtains a plurality of two-dimensional arrays of superimposition coefficients corresponding to the reduced grayscale images according to the multiple sets of energy gradient groups obtained in step 24, respectively corresponding to the reduced grayscale images. Each two-dimensional array of overlapping coefficients includes a plurality of overlapping coefficients respectively corresponding to the pixels of the corresponding gray scale reduced image. For example, if the resolution of each grayscale reduced image is 3×3, because there are 9 pixels in total, the number of the overlap coefficients is 9, and the two-dimensional array of the overlap coefficients is 3× 3 arrays.

With reference to FIG. 4, step 25 includes sub-steps 251 to 255, and the sub-steps included in step 25 are described below.

In step 251, for each pixel position of the reduced grayscale images, the processing unit 12 obtains one of the maximum energy gradients of all the energy gradients corresponding to the pixel position and one of the energy gradients of all the energy gradients corresponding to the pixel position. The sum of the gradients.

In step 252, for each pixel position of each grayscale reduced image, the processing unit 12 determines whether the maximum energy gradient corresponding to the pixel position of the grayscale reduced image is greater than a threshold value. For each pixel position of each grayscale reduced image, when it is determined that the maximum energy gradient corresponding to the pixel position of the grayscale reduced image is greater than the threshold value, the process proceeds to step 253; and when the grayscale is determined When the maximum energy gradient corresponding to the pixel position of the reduced image is not greater than the threshold value, the process proceeds to step 254.

。值得注意的是，在本實施例中，

，

為位於該灰階縮小圖像的該像素位置之像素的鄰域像素之數量，在其他實施方式中，

亦可為其他值，例如

、

，及

，但不以此為限。 The threshold value is as follows: Threshold value = EG _max * P %, where EG _max is a positive number,

. It is worth noting that in this embodiment,

,

Is the number of neighboring pixels of the pixel located at the pixel position of the grayscale reduced image. In other embodiments,

Can also be other values, such as

,

,and

, But not limited to this.

In step 253, for each pixel position of each gray scale reduced image, the processing unit 12 obtains a corresponding gray scale reduction according to the sum of energy gradients and energy gradients corresponding to the pixel position of the gray scale reduced image The overlap coefficient of the pixel position of the image. It is worth noting that, in this embodiment, the overlap coefficient is the energy gradient corresponding to the pixel position divided by the sum of the energy gradient corresponding to the pixel position, but it is not limited to this.

In step 254, for each pixel position of each gray-scale reduced image, the processing unit 12 obtains the overlap coefficient according to the number of the gray-scale reduced images. It is worth noting that, in this embodiment, the superimposition factor is 1 divided by the number of the gray scale reduced images, but it is not limited to this.

In step 255, for each gray-scale reduced image, the processing unit 12 obtains a superimposed coefficient corresponding to the gray-scale reduced image according to the overlap coefficients corresponding to all pixel positions of the gray-scale reduced image. Dimensional array.

Refer to Figure 5 for collocation. For example, suppose there are 3 reduced grayscale images, and the resolution of each reduced grayscale image is 3×3, so there are (1,1), (1,2), (1) ,3), (2,1), (2,2), (2,3), (3,1), (3,2), and (3,3) 9 pixel positions.

，若 P=1，則該門檻值=1300.5，由於像素位置(1,1)對應的最大能量梯度大於該門檻值，故對應第1張灰階縮小圖像的像素位置(1,1)之疊合係數為1423/2887=0.493，對應第2張灰階縮小圖像的像素位置(1,1)之疊合係數為867/2887=0.3，對應第3張灰階縮小圖像的像素位置(1,1)之疊合係數為597/2887=0.207。要特別注意的是，該等灰階縮小圖像的像素位置(1,1)對應的疊合係數總合為1，故對應第3張灰階縮小圖像的像素位置(1,1)之疊合係數亦可由1-0.493-0.3=0.207的方式獲得。 If the energy gradients corresponding to the pixel position (1,1) of these grayscale reduced images are 1423, 867, and 597, respectively, the maximum energy gradient corresponding to the pixel position (1,1) is 1423, and the pixel position (1,1) ) Corresponds to the sum of the energy gradients of 2887. Taking 4-connectivity as the standard, the number of neighboring pixels of the pixel at the pixel position (1,1) is 2, so

, If P =1, the threshold value = 1300.5. Since the maximum energy gradient corresponding to the pixel position (1,1) is greater than the threshold value, it corresponds to the pixel position (1,1) of the first grayscale reduced image The overlap coefficient is 1423/2887=0.493, which corresponds to the pixel position (1,1) of the second grayscale reduced image, and the overlap coefficient is 867/2887=0.3, which corresponds to the pixel position of the third grayscale reduced image The overlap coefficient of (1,1) is 597/2887=0.207. It should be noted that the sum of the overlap coefficients corresponding to the pixel position (1,1) of the reduced grayscale image is 1, so it corresponds to the pixel position (1,1) of the third grayscale reduced image. The overlap coefficient can also be obtained by the method of 1-0.493-0.3=0.207.

，若 P=1，則該門檻值=1950.75，由於像素位置(1,1)對應的最大能量梯度不大於該門檻值，故對應第1張灰階縮小圖像的像素位置(1,2)之疊合係數為1/3，對應第2張灰階縮小圖像的像素位置(1,2)之疊合係數為1/3，對應第3張灰階縮小圖像的像素位置(1,2)之疊合係數為1/3。 If the energy gradients corresponding to the pixel positions (1,2) of the reduced grayscale images are 895, 1023, and 485, respectively, the maximum energy gradient corresponding to the pixel position (1,2) is 1023, and the energy corresponding to the pixel position The sum of gradients is 2403. Taking 4 connectivity as the standard, the number of neighboring pixels of the pixel at pixel position (1,2) is 3, so

, If P =1, the threshold value = 1950.75, since the maximum energy gradient corresponding to the pixel position (1,1) is not greater than the threshold value, it corresponds to the pixel position (1,2) of the first grayscale reduced image The overlap coefficient is 1/3, and the overlap coefficient corresponding to the pixel position (1,2) of the second grayscale reduced image is 1/3, which corresponds to the pixel position (1, 2) The overlap coefficient is 1/3.

In step 26, for each overlap coefficient of each overlap coefficient two-dimensional array, the processing unit 12 obtains the coefficient average value of the overlap coefficient and the adjacent overlap coefficient.

In step 27, for each two-dimensional array of overlapping coefficients, the processing unit 12 obtains a smoothed overlapping coefficient two according to the average value of all the coefficients corresponding to the two-dimensional array of overlapping coefficients obtained in step 26. Dimensional array.

It should be noted that, in this embodiment, for each two-dimensional array of superimposed coefficients, the processing unit 12 replaces its corresponding superimposed coefficient with the average value of the obtained coefficients to obtain a smoothed superimposed system. For example, suppose that the overlap coefficient corresponding to the first grayscale reduced image is the overlap coefficient of the first column and the first row of the two-dimensional array (that is, the overlap coefficient corresponding to the first grayscale reduced image The overlap coefficient at the pixel position (1,1) is 0.493, and if 4 is connected as the standard, the adjacent overlap coefficient is the first column of the two-dimensional array of overlap coefficients corresponding to the first grayscale reduced image The overlap coefficient of the second row, for example 1/3, and the overlap coefficient of the second column and the first row of the two-dimensional array corresponding to the first grayscale reduced image, for example 0.4, can be obtained Corresponding to the first gray-scale reduced image, the superimposed coefficient corresponding to the superimposed coefficient in the first column and the first row of the two-dimensional array is 0.409, so it corresponds to the smoothing of the first gray-scale reduced image The value of the first column and the first row of the superposition coefficient two-dimensional array is 0.409.

It should be noted again that in this embodiment, the processing unit 12 first obtains the corresponding coefficient average value for each overlap coefficient, and then obtains a two-dimensional array of smoothed overlap coefficients from the coefficient average value. In other implementations, In the manner, the processing unit 12 may also perform a Gaussian blur operation on the two-dimensional arrays of overlapping coefficients to obtain a plurality of smoothed two-dimensional arrays of overlapping coefficients respectively corresponding to the two-dimensional arrays of overlapping coefficients. , Not limited to this.

In step 28, the processing unit 12 amplifies the two-dimensional array of smoothed superimposition coefficients to obtain a plurality of magnified superimposed coefficients corresponding to the two-dimensional array of smoothed superimposition coefficients. The size of the two-dimensional array of the magnified superimposed coefficient two-dimensional array is the same as the resolution of the image after the image processing.

It is worth noting that, in this embodiment, the processing unit 12 uses a neighbor correlation interpolation method to amplify the two-dimensional array of overlapping coefficients, but it is not limited to this.

It should be noted again that in other embodiments, steps 26 and 27 may not be performed, so that in step 28, the processing unit 12 enlarges the two-dimensional array of overlapping coefficients.

In step 29, the processing unit 12 superimposes the processed images according to the two-dimensional arrays of magnified superimposition coefficients to obtain a superimposed image.

Referring to FIG. 6 in conjunction, step 29 includes sub-steps 291 to 293. The sub-steps included in step 29 are described below.

In step 291, for each image after image processing, the processing unit 12 superimposes a pixel array including all pixel values of the image after image processing with the corresponding magnification of the image after image processing. The two-dimensional array of coefficients is multiplied to obtain a superimposed pixel array.

In step 292, the processing unit 12 superimposes all the superimposed pixel arrays obtained in step 291 to obtain a superimposed pixel array.

In step 293, the processing unit 12 obtains the superimposed image according to the superimposed pixel array.

It should be particularly noted that in other embodiments, step 22 may not be performed, and the processing unit 12 directly performs steps 23 to 29 with the captured images, so that in step 23, the processing unit 12 performs the captured images. The image is gray-scaled and reduced. In step 28, the array size of the two-dimensional array of magnification and superimposition coefficients is the same as the resolution of the captured images, and in step 29, the processing unit 12 is The captured images are superimposed.

It should be noted again that in other embodiments, the processing unit 12 may also perform image processing on the superimposed image, such as image sharpening, to obtain a sharpened superimposed image with clearer feature details. Like, but not limited to it.

In summary, the semiconductor device image superimposing method of the present invention superimposes the photographed images with different characteristics through the processing unit 12, so that the obtained superimposed image can reflect the semiconductor The difference in height and reflectance of the surface of the element. In addition, the processing unit 12 obtains the two-dimensional arrays of overlap coefficients according to the energy gradient groups, and then smoothes and enlarges the two-dimensional arrays of overlap coefficients Finally, the superimposed image is obtained according to the two-dimensional array of magnified superimposition coefficients, which can prevent the phenomenon of blocky distribution in the superimposed image, so the objective of the invention can be achieved.

However, the above are only examples of the present invention. When the scope of implementation of the present invention cannot be limited by this, all simple equivalent changes and modifications made in accordance with the scope of the patent application of the present invention and the content of the patent specification still belong to Within the scope covered by the patent of the present invention.

1: Image overlay device 11: Image capture unit 12: Processing unit 21~29: Steps 221~223: Steps 251~255: Step 291~293: Steps

Other features and effects of the present invention will be clearly presented in the embodiments with reference to the drawings, in which: 1 is a block diagram illustrating an image superimposing device for implementing an embodiment of the semiconductor device image superimposing method of the present invention; FIG. 2 is a flowchart illustrating the embodiment of the semiconductor device image superimposing method of the present invention; Figure 3 is a flowchart to assist in explaining the sub-steps of step 22 in Figure 2; Figure 4 is a flowchart to assist in explaining the sub-steps of step 25 in Figure 2; Figure 5 is a schematic diagram illustrating 3 reduced gray scale images; and FIG. 6 is a flowchart to assist in explaining the sub-steps of step 29 in FIG. 2.

21~29: Steps

Claims (17)

A method for superimposing images of a semiconductor element, which is suitable for superimposing an image of a semiconductor element, is implemented by an image superimposing device, the image superimposing device includes an image capturing unit, and an electrical connection to the image A processing unit of a photographing unit, the method includes the following steps: (A) by the image photographing unit, photographing the semiconductor device according to a plurality of different photographing parameter data to obtain multiple photographs corresponding to the photographing parameter data and having the same Resolution photographed images; (B) by the processing unit, the photographed images are subjected to image processing to obtain a plurality of processed images corresponding to the photographed images. Step (B) includes The following sub-steps: (B-1) using the processing unit, for each captured image, smoothing the captured image to obtain a smoothed image, (B-2) using the processing unit, For each captured image, subtract the corresponding smoothed image from the captured image to obtain an edge detail image, and (B-3) With the processing unit, for each captured image, Superimpose the captured image with the corresponding edge detail image to obtain a processed image corresponding to the captured image; and (C) through the processing unit, the processed image The images are superimposed to obtain a superimposed image. The semiconductor device image superimposing method according to claim 1, wherein step (C) includes the following sub-steps: (C-1) using the processing unit to grayscale the captured images And reduction to obtain multiple grayscale reduced images with the same resolution corresponding to the captured images; (C-2) through the processing unit, for each grayscale reduced image, according to the grayscale Obtain an energy gradient group including a plurality of energy gradients respectively corresponding to the pixel positions of the pixels of the gray-scale reduced image, and the pixel positions correspond to the pixels; (C- 3) With the processing unit, according to the multiple sets of energy gradient groups corresponding to the reduced grayscale images obtained in step (C-2), multiple overlap coefficients corresponding to the reduced grayscale images are obtained. Two-dimensional array, each two-dimensional array of superimposition coefficients includes a plurality of superimposition coefficients corresponding to the pixels of the corresponding gray scale reduced image; (C-4) through the processing unit, the superimposition system The two-dimensional array is enlarged to obtain a plurality of two-dimensional arrays of enlarged superimposition coefficients corresponding to the two-dimensional arrays of superimposed coefficients. Wait for the resolution of the captured images to be the same; and (C-5) by the processing unit, superimpose the captured images according to the two-dimensional array of the magnified superimposition coefficients to obtain the superimposed image picture. The semiconductor device image superimposing method according to claim 2, wherein, in step (C-1), the processing unit gray-scales the captured images, and uses a neighboring correlation interpolation The method reduces the side lengths of the captured images after gray-scale to the same proportion. The semiconductor device image overlay method according to claim 2, wherein, in step (C-2), for each gray-scale reduced image, the energy gradient group corresponding to the gray-scale reduced image includes the The iso-energy gradient is as follows:

Where ( x _i , y _j ) is the pixel position of one of the target pixels in the grayscale reduced image, EG ( x _i , y _j ) is the energy gradient corresponding to the pixel position of the target pixel, ( x _p , y _q ) is the pixel position of the neighboring pixel of the target pixel, pixel ( x _i , y _j ) is the pixel value of the target pixel, pixel ( x _p , y _q ) is the neighboring pixel of the target pixel The pixel value. The semiconductor device image overlay method according to claim 2, wherein, in step (C-2), for each gray-scale reduced image, the energy gradient group corresponding to the gray-scale reduced image includes the The iso-energy gradient is as follows:

Where ( x _i , y _j ) is the pixel position of one of the target pixels in the grayscale reduced image, EG ( x _i , y _j ) is the energy gradient corresponding to the pixel position of the target pixel, ( x _p , y _q ) is the pixel position of the neighboring pixel of the target pixel, pixel ( x _i , y _j ) is the pixel value of the target pixel, pixel ( x _p , y _q ) is the neighboring pixel of the target pixel The pixel value. The semiconductor device image overlay method according to claim 2, wherein, in step (C-2), for each gray-scale reduced image, the energy gradient group corresponding to the gray-scale reduced image includes the The iso-energy gradient is as follows:

Where ( x _i , y _j ) is the pixel position of one of the target pixels in the grayscale reduced image, EG ( x _i , y _j ) is the energy gradient corresponding to the pixel position of the target pixel, ( x _p , y _q ) is the pixel position of the neighboring pixel of the target pixel, pixel ( x _i , y _j ) is the pixel value of the target pixel, pixel ( x _p , y _q ) is the neighboring pixel of the target pixel The pixel value. The semiconductor device image overlay method according to claim 2, wherein, in step (C-2), for each gray-scale reduced image, the energy gradient group corresponding to the gray-scale reduced image includes the The iso-energy gradient is as follows:

Where ( x _i , y _j ) is the pixel position of one of the target pixels in the grayscale reduced image, EG ( x _i , y _j ) is the energy gradient corresponding to the pixel position of the target pixel, ( x _p , y _q ) is the pixel position of the neighboring pixel of the target pixel, pixel ( x _i , y _j ) is the pixel value of the target pixel, pixel ( x _p , y _q ) is the neighboring pixel of the target pixel The pixel value of, N _{i , j is} the number of neighboring pixels of the target pixel. The semiconductor device image overlay method according to claim 2, wherein the step (C-3) includes the following sub-steps: (C-3-1) through the processing unit, for the grayscale reduced image For each pixel position, obtain one of the largest energy gradients of all the energy gradients corresponding to the pixel position and the sum of one of the energy gradients of all energy gradients corresponding to the pixel position; (C-3-2) With the processing unit, for each For each pixel position of a reduced gray scale image, determine whether the maximum energy gradient corresponding to the pixel position of the reduced gray scale image is greater than a threshold value, the threshold value is as follows: threshold value = EG _max * P %, EG _max is a positive number, 0

P

100; (C-3-3) With the processing unit, for each pixel position of each grayscale reduced image, when it is determined that the maximum energy gradient corresponding to the pixel position of the grayscale reduced image is greater than the threshold Value, according to the sum of the energy gradient and the energy gradient corresponding to the pixel position of the gray-scale reduced image, obtain the superimposition coefficient corresponding to the pixel position of the gray-scale reduced image; (C-3-4) by The processing unit, for each pixel position of each gray-scale reduced image, when it is determined that the maximum energy gradient corresponding to the pixel position of the gray-scale reduced image is not greater than the threshold value, according to the gray-scale reduced images The number of images to obtain the superimposition coefficient; and (C-3-5) by the processing unit, for each gray scale reduced image, the superimposed coefficient corresponding to all pixel positions of the image is reduced according to the gray scale, Obtain a two-dimensional array of superimposition coefficients corresponding to the gray scale reduced image. The semiconductor element image superimposing method according to claim 8, wherein, in step (C-3-2), EG _max =255 ² * N _{i , j} , N _{i , j} is the gray scale reduced image The number of neighboring pixels of the pixel at the pixel position of the image. The semiconductor device image superimposing method according to claim 8, wherein, in step (C-3-2), EG _max =255* N _{i , j} , N _{i , j} is a reduced image located in the gray scale The number of neighboring pixels of the pixel at the pixel position. The semiconductor element image superimposing method according to claim 8, wherein, in step (C-3-2),

, N _{i , j} are the number of neighboring pixels of the pixel located at the pixel position of the reduced gray scale image. The semiconductor element image superimposing method according to claim 8, wherein, in step (C-3-2), EG _max =255. The semiconductor element image superimposing method according to claim 2, wherein, in step Between step (C-3) and step (C-4), the following sub-steps are also included: (C-6) by the processing unit, for each superimposed coefficient of the two-dimensional array of superimposed coefficients, obtain The coefficient average value of the overlap coefficient and its adjacent overlap coefficient; and (C-7) by the processing unit, for each two-dimensional array of overlap coefficients, all the values obtained in step (C-6) Corresponding to the coefficient average value of the two-dimensional array of overlapping coefficients, a smoothed two-dimensional array of overlapping coefficients is obtained; wherein, in step (C-4), the smoothing obtained in step (C-7) The two-dimensional array of overlapping coefficients is enlarged to obtain the two-dimensional array of enlarged overlapping coefficients. The semiconductor device image overlay method according to claim 2, wherein, between step (C-3) and step (C-4), the following sub-steps are further included: (C-8) through the processing unit, Perform a Gaussian blur operation on the two-dimensional arrays of overlapping coefficients to obtain a plurality of smoothed two-dimensional arrays of overlapping coefficients respectively corresponding to the two-dimensional arrays of overlapping coefficients; wherein, in step (C-4 In ), the two-dimensional array of smoothed superimposition coefficients obtained in step (C-8) is enlarged to obtain the two-dimensional array of enlarged superimposed coefficients. The semiconductor device image superimposing method according to claim 2, wherein, in step (C-4), the processing unit uses a neighboring correlation interpolation method for the two-dimensional array of superimposed coefficients enlarge. The semiconductor device image superimposing method according to claim 2, wherein step (C-5) includes the following sub-steps: (C-5-1) by the processing unit, for each captured image, a The pixel array including all the pixel values of the captured image is multiplied by the two-dimensional array of magnification superimposed coefficients corresponding to the captured image to obtain a superimposed pixel array; (C-5-2) by this processing Unit, superimpose all the superimposed pixel arrays obtained in step (C-5-1) to obtain a superimposed pixel array; and (C-5-3) by the processing unit, obtain a superimposed pixel array according to the superimposed pixel array The superimposed image. The semiconductor device image superimposing method according to claim 1, after step (C), further includes the following steps: (D) by the processing unit, perform an image sharpening operation on the superimposed image to Obtain a sharpened superimposed image.

Images