Multi-angle illumination image synthesis method and device

Abstract

The invention discloses a multi-angle illumination image synthesis method and device, and the method comprises the steps: obtaining a plurality of illumination images of a same sample at different illumination visual angles, taking the center of the illumination image as an observation center, carrying out the difference calculation of two illumination images of which the illumination visual angles are located in a straight line where the observation center is located, and obtaining a plurality of difference images; respectively determining enhancement multiples according to the brightness of the plurality of original illumination images, enhancing the difference image according to the determined enhancement multiples, and performing interpolation calculation on the obtained difference enhanced image to obtain a longitude image; obtaining an illumination image of a top view angle or calculating and processing a plurality of original illumination images to synthesize the illumination image of the top view angle; and performing interpolation calculation on the obtained longitude image and the illumination image of the top view angle to obtain an illumination image with given longitude and latitude. And an image observed from any angle on the hemispherical surface above the sample can be obtained.

Description

Technical Field

[0001] The invention relates to the field of image processing, and in particular to a multi-angle illumination image synthesis method and device. Background Art

[0002] With the rapid development of software and hardware technology, automatic microscopes have been increasingly used in industrial inspection. As an important function in industrial inspection, multi-angle lighting helps to observe and detect the texture and defects on the surface of the sample at a specific viewing angle. However, limited by the hardware structure, automatic microscopes can only achieve lighting at a few fixed angles. To achieve lighting at any viewing angle, it is necessary to synthesize it through algorithms and software. One approach is to first illuminate the area to be observed of the sample at several fixed angles and take the corresponding images (called source images), and then use algorithms to synthesize them to obtain images "illuminated" from any angle in the hemispherical space.

[0003] The image quality of arbitrary-angle illumination depends on the clarity of the source image, the accuracy of the light source angle, and the accuracy of the synthesis algorithm. The clarity of the original image and the accuracy of the light source angle can be solved by adjusting the microscope's own structure and magnification, while the accuracy of the synthesis algorithm means that the synthesized image (i.e., the "synthetic image") must be consistent with the expected "real image" under the corresponding illumination angle, which requires the establishment of a correct mathematical model for multi-angle illumination.

[0004] Currently, image synthesis is usually completed by image interpolation. For example, linear interpolation between two images can be used to simulate the smooth movement of light sources. In addition, the image often needs to be enhanced before synthesis, which can be achieved by performing point processing, histogram equalization, image sharpening, high-pass filtering, gamma correction, etc.

[0005] However, the linear interpolation between the two images shows the translation of the light source, which is inconsistent with multi-angle lighting because the latter simulates the movement of the light source around a hemispherical surface. The image generated by the light source from any angle of view on the hemispherical surface should have a bright side and a dark side, where the bright side is the side that the user focuses on. If you need to observe the dark side, you need to adjust the light source to the opposite angle of view to turn the original dark side into a bright side.

[0006] In addition, for images with uneven illumination, point processing and histogram processing of the image cannot effectively enhance the bright side of the image or may produce unwanted enhancement effects on the dark side. Sharpening the image will change the global texture details of the image to a certain extent, and high-pass filtering in the frequency domain can easily filter out effective texture details. The gamma correction method can enhance the contrast of the image, but for images with uneven angle illumination, this method is troubled by oversaturation and undersaturation problems. The fuzzy-based method uses fuzzy theory to judge the brightness of the image and distinguish between low-brightness and high-brightness areas and process them separately. However, the pixels of the bright side and the dark side in the image are often cross-mixed. The intuitive understanding is that the side facing the light source is the bright side, and the side facing away from the light source is the dark side. If only the low-brightness area and the high-brightness area are processed separately, it is still impossible to accurately enhance the bright side. Summary of the invention

[0007] The main purpose of the present invention is to overcome the problem that in current automatic microscope observations, samples can only be observed at a few fixed illumination angles, and to propose a multi-angle illumination image synthesis method and device that can obtain images observed at any angle on the hemispherical surface above the sample.

[0008] The present invention adopts the following technical solution:

[0009] A multi-angle illumination image synthesis method, comprising:

[0010] Acquire a plurality of illumination images of the same sample at different illumination angles, take the center of the illumination image as the observation center, and perform differential calculation on two illumination images whose illumination angles are on a straight line where the observation center is located to obtain a plurality of differential images;

[0011] Determining enhancement multiples according to the brightness of the original illumination images respectively, enhancing the difference image according to the determined enhancement multiples, and performing interpolation calculation on the obtained difference enhancement image to obtain a longitude image;

[0012] Acquire an illumination image from a top perspective or calculate and process a plurality of original illumination images to synthesize an illumination image from a top perspective;

[0013] The obtained longitude image and the illumination image of the top viewing angle are interpolated to obtain an illumination image of given longitude and latitude.

[0014] The pixels of the two illumination images whose illumination angles are on the straight line where the observation center is located are differentially calculated, specifically including the following:

[0015] D TL =I TL -I BR ={d i,j,c,i=1Kh,j=1Kw,c=1,2,3}

[0016] d i,j,c =d i,j,c ,d i,j,c >0

[0017] d i,j,c =0,d i,j,c ≤0

[0018] Among them, I TL and I BR The illumination angle is the two illumination images of the straight line where the observation center is located, D TL is the illumination image I TL The difference image, d i,j,c is the difference image D TL The matrix elements of , i, j, c are the row number, column number and channel number of the matrix respectively, and h is the difference image D TL The height of w is the difference image D TL Width.

[0019] The enhancement factor is determined according to the brightness of the original illumination images: if the brightness of the original illumination image is dark, the enhancement factor of the differential image corresponding to the original illumination image is increased; if the brightness of the original illumination image is bright, the enhancement factor of the differential image corresponding to the original illumination image is reduced; the brightness of the original image is calculated according to the mean of the histogram.

[0020] The method further includes determining whether the differential enhanced image is overexposed according to the number of pixels of the differential enhanced image, and if so, adjusting the enhancement multiple of the differential image until the differential enhanced image is not overexposed, thereby obtaining the enhancement multiple.

[0021] The relationship between the differential enhanced image and the enhancement multiple is as follows:

[0022]

[0023] Among them I TL ,I BR ,I TR ,I BL is the original illumination image under four different illumination angles, α is the determined enhancement multiple, I' TL , I' TR , I' BL , I' BR are the original illumination images I under different illumination angles respectively. TL ,I BR ,I TR ,I BL The corresponding difference enhanced image.

[0024] The obtained differential enhanced image is interpolated to obtain the longitude image Specifically:

[0025]

[0026] Among them, I' TL , I' TR , I' BR , I' BL are differentially enhanced images corresponding to the original illumination images under four different illumination viewing angles, is the longitude.

[0027] The original illumination images are averaged and then enhanced to obtain the illumination image of the top perspective, as follows:

[0028]

[0029] Among them, I top is the illumination image of the top view, I TL ,I BR ,I TR ,I BL are the original illumination images under four different illumination perspectives.

[0030] The obtained longitude image is interpolated with the synthesized illumination image of the top perspective to obtain an illumination image of a given longitude and latitude. The details are as follows:

[0031]

[0032] in The latitude is 0 and the longitude is The angle illumination image, I top is the illumination image of the top viewing angle, with 90-θ as the dimension.

[0033] A multi-angle illumination image synthesis device, comprising:

[0034] A difference module, which obtains a plurality of illumination images of the same sample at different illumination angles, takes the center of the illumination image as the observation center, and performs difference calculation on two illumination images whose illumination angles are on a straight line where the observation center is located to obtain a plurality of difference images;

[0035] A first interpolation module, which determines enhancement multiples according to the brightness of the original illumination images, enhances the difference image according to the determined enhancement multiples, and performs interpolation calculation on the obtained difference enhancement image to obtain a longitude image;

[0036] A synthesis module, which calculates and processes the original plurality of illumination images to synthesize an illumination image of a top perspective;

[0037] The second interpolation module performs interpolation calculation on the obtained longitude image and the synthesized illumination image of the top perspective to obtain an illumination image of given longitude and latitude.

[0038] It can be seen from the above description of the present invention that, compared with the prior art, the present invention has the following beneficial effects:

[0039] In the present invention, differential enhancement is applied to establish a light and dark surface model of an image, and then two interpolations are performed to obtain an illumination image from any angle on the hemispherical surface. This reduces the requirements for the number of optical angle illumination components of imaging equipment such as industrial measuring microscopes, and reduces costs. This enhances the all-round observation of sample texture and defects by imaging equipment such as industrial measuring microscopes.

[0040] In the present invention, the image quality is improved by differential enhancement of the image; the calibration work of imaging equipment such as industrial measuring microscopes is reduced. For example, there is no need to set different light intensities for different samples, and image enhancement can be performed adaptively, thereby improving work efficiency.

[0041] In the present invention, the light and dark side model of the image is effectively established by differencing the illumination image, so as to specifically enhance only the bright side of the image. By analyzing the relationship between the image histogram index and the enhancement multiple, the image enhancement degree is set in a reasonable range, so that the image enhancement has a significant effect but will not be overexposed.

[0042] In the present invention, an effective synthetic top lighting image is enhanced by coefficients based on the image average, so that the brightness and contrast enhancement are obvious; several original lighting images are placed in a spherical coordinate system, and weighted interpolation is performed according to longitude and latitude to meet the needs of lighting images at different longitudes and latitudes. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a flow chart of the method of the present invention;

[0044] Figure 2 are the original illumination images of the four illumination perspectives;

[0045] Figure 3 is the image after diagonal difference;

[0046] Figure 4 The illumination images of two samples with darker brightness and brighter brightness under a certain illumination angle;

[0047] Figure 5 is the relationship between the mean pixel value and the differential enhancement multiple;

[0048] Figure 6 The relationship between the point increase multiple and the differential enhancement multiple for a pixel value of 255;

[0049] Figure 7 The illumination images are from the top perspective with different enhancement factors;

[0050] Figure 8 Schematic diagram of image interpolation at different longitude angles;

[0051] Fig. 9 Illumination images at different longitude angles generated for interpolation;

[0052] Fig.10 Schematic diagram of generating illumination images at different longitudes and latitudes by interpolation of spherical coordinate system; Fig.11 Illumination images at different latitude angles generated for interpolation;

[0053] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. DETAILED DESCRIPTION

[0054] The present invention is further described below through specific implementation modes.

[0055] See also Figure 1 , a multi-angle illumination image synthesis method, comprising:

[0056] 1) Obtain several illumination images of the same sample at different illumination angles, take the center of the illumination image as the observation center, and perform differential calculation on two illumination images whose illumination angles are on the straight line where the observation center is located to obtain several differential images.

[0057] In the present invention, the different lighting viewing angles can be multiple, including two, three, four or even more. Taking four lighting viewing angles as an example, Figure 1 , respectively use the light sources in the upper left corner, upper right corner, lower left corner, and lower right corner to illuminate the same sample, and obtain the original illumination image I TL ,I BR ,I TR ,I BL , which are the illumination image of the upper left corner, the illumination image of the upper right corner, the illumination image of the lower left corner, and the illumination image of the lower right corner respectively.

[0058] Take the center of the illumination image as the observation center (the end point of the line of sight), and take two illumination images that are on the straight line of the observation center and opposite to each other (for example, the upper left to the lower right, the lower left to the upper right). The bright side and the dark side of the image are distinguished by difference. The differenced image is as follows: Figure 2 .

[0059] Figure 2If the pixel value difference is positive, the information is retained, and if the pixel value difference is negative, it is set to zero. TL and I BR For example, the illumination image I TL The image difference calculation formula is as follows:

[0060] D TL =I TL -I BR ={d i,j,c ,i=1Kh,j=1Kw,c=1,2,3}

[0061] d i,j,c =d i,j,c ,d i,j,c >0

[0062] d i,j,c =0,d i,j,c ≤0 (1)

[0063] Among them, I TL and I BR D is the two illumination images of the straight line where the illumination angle is located at the observation center. TL I is the illumination image TL The difference image, d i,j,c is the difference image D TL The matrix elements of , i, j, c are the row number, column number and channel number of the matrix respectively, and h is the difference image D TL The height of w is the difference image D TL Similarly, the illumination image I BR The difference image is D BR =I BR -I TL , lighting image I TR The difference image is D TR =I TR -I BL , lighting image I BL The difference image is D BL =I BL -I TR .

[0064] Since the bright surface is the area that is more emphasized in sample observation, it is necessary to enhance the bright surface.

[0065] 2) Determine enhancement multiples according to the brightness of the original illumination images, enhance the difference image according to the determined enhancement multiples, and perform interpolation calculation on the obtained difference enhanced image to obtain the longitude image.

[0066] In this step, the enhancement factor is determined according to the brightness of several original illumination images: if the brightness of the original illumination image is dark, the enhancement factor of the differential image corresponding to the original illumination image is increased; if the brightness of the original illumination image is bright, the enhancement factor of the differential image corresponding to the original illumination image is reduced; the brightness of the original image is calculated according to the mean of the histogram.

[0067] For the determination of the enhancement multiple, the difference image of two samples (a PCB board and a 5-cent coin) with dark and bright brightness under a certain lighting angle is selected as an example, Figure 3 As shown. Binarization threshold processing is performed on the difference image, and the elements with pixel values ​​greater than 0 in the image matrix of the three RGB channels in the difference image are set to 1, and the calculation formula of the image label matrix A representing the bright surface is obtained as follows:

[0068]

[0069] where a i,j,c is the element of the image labeling matrix A, d i,j,c is the element in the difference image D in formula (1), i, j, c are the row number, column number and channel number of the matrix respectively, h is the height of image A, and w is the width of image A.

[0070] The matrix A is multiplied by the original illumination image I of the illumination angle to obtain the bright surface part I of the illumination image a , the calculation formula is as follows:

[0071] I a =I·A (3)

[0072] Subtract the image matrix A from the matrix with the same size and all elements with the value of 1 to get the image label matrix B representing the dark side. Multiply the matrix B with the original illuminated image I under the illumination vision to get the dark side part I of the illuminated image. b , the calculation formula is as follows:

[0073] b i,j,c =1-a i,j,c (4)

[0074] I b =I·B (5)

[0075] where b i,j,c is the element of matrix B, i, j, c are the row number, column number and channel number of matrix B respectively.

[0076] The relationship between the difference enhanced image and the difference enhancement multiple is as follows:

[0077]

[0078] The histogram mean of the bright part and the dark part of the differential enhanced image is calculated respectively, and the pixel value mean of the bright part is

[0079] The relationship between the differential enhancement factor and the differential enhancement factor is as follows (7):

[0080]

[0081] in It represents the number of points with pixel value i in the histogram statistics of the bright part of the image c channel after differential enhancement n times, a i, j ,c is the element of the image label matrix A in formula (2), h and w are the height and width of image A respectively. The relationship between the mean pixel value and the differential enhancement factor expressed in formula (7) is plotted as follows: Figure 4 The left picture shows the curve of the sample PCB board, and the right picture shows the curve of the 5-cent coin. The red curve is the relationship curve of the bright side, and the blue curve is the relationship curve of the dark side.

[0082] After experiments, the enhancement multiple corresponding to the average value of 125 is the adaptive enhancement multiple, which has a more obvious enhancement effect and is not easy to overexpose. At this time, the differential enhancement multiple of the dark PCB board image is 1.1 times, and the differential enhancement multiple of the bright 5-cent coin image is 0.4 times. In this way, the differential enhancement multiple can be adaptively adjusted to a reasonable value for different samples.

[0083] As the image is enhanced, the number of points with a pixel value of 255 will increase. To prevent overexposure, a range of 255 pixel points of the image needs to be set. If it exceeds this range, it is considered overexposed. Figure 5 The relationship curve between the increase of 255 pixels in the bright area and the differential enhancement multiple is shown. The sample on the left is a PCB board, and the sample on the right is a 5-cent coin. If the multiple is limited too low, the enhancement effect is not obvious. After experiments, it is generally set to 6 times, but not limited to this. At this time, the upper limit enhancement multiple of the PCB board is 1.4 times, and the upper limit enhancement multiple of the 5-cent coin is 0.8 times.

[0084] The relationship between the point increase factor and the difference enhancement factor for the bright part of the image with a pixel value of 255 is as follows (8):

[0085]

[0086] in is the number of points where the pixel value of channel c of the bright part of the image after differential enhancement by n times is 255. The number of points where the pixel value of the c channel of the bright part of the image without differential enhancement is 255.

[0087] This step also includes determining whether the differential enhanced image is overexposed according to the number of pixels of the differential enhanced image. If so, adjusting the enhancement multiple of the differential image until the differential enhanced image is not overexposed, thereby obtaining the enhancement multiple.

[0088] Specifically, whether the image is overexposed is determined based on whether the ratio between the pixel points of the differential enhanced image and the enhancement multiple is a set value or within a set range. If the enhanced image is overexposed, the differential enhancement multiple is adjusted so that the image is not overexposed. For example, if the point increase multiple of the 255 pixel value exceeds 6 times after image enhancement, the differential enhancement multiple is reduced so that the point increase multiple of the 255 pixel value is 6 times, thereby limiting the differential enhancement multiple to a reasonable range.

[0089] The present invention effectively establishes a light and dark side model of an image by performing image difference on a pair of face images, thereby specifically enhancing only the bright side of the image. By analyzing the relationship between the image histogram index and the enhancement multiple, the image enhancement degree is set within a reasonable range, so that the image enhancement has an obvious effect but will not be overexposed.

[0090] After determining the enhancement multiple, the difference enhancement image is calculated for interpolation to generate an image with any viewing angle. The relationship between the difference enhancement image and the enhancement multiple is as follows:

[0091]

[0092] Among them I TL ,I BR ,I TR ,I BL is the original illumination image under four different illumination angles, α is the determined enhancement factor, I' TL , I' TR , I' BL , I' BR They are the original illumination images I under different illumination perspectives. TL ,I BR ,I TR ,I BL The corresponding difference enhanced image.

[0093] Based on the difference enhancement of multiple lighting angles, take four lighting angles as an example. These four angles are on the same plane, that is, at the same latitude. The default latitude is 0, such as Figure 7 shown.

[0094] With upper left corner The calculation of the angle illumination image is as follows:

[0095]

[0096] Taking the longitude of due north as 0 degrees, as the longitude increases in the clockwise direction, the obtained differential enhanced image is interpolated to obtain the longitude Longitude image from 0 to 360 degrees Specifically:

[0097]

[0098] Among them, I′ TL , I′ TR , I′ BR , I′ BL They are the difference enhanced images corresponding to the original illumination images under four different illumination angles. is the longitude. The effect after interpolation is as follows Figure 8 , Figure 8 The light source angles in the middle left and right pictures are longitude 0 degrees and longitude 70 degrees respectively, and the latitude is 0 degrees.

[0099] 3) Obtain an illumination image from a top perspective or calculate and process several original illumination images to synthesize an illumination image from a top perspective.

[0100] In this step, if there is an original suitable top-view illumination image, it can be directly applied to the latitude image interpolation in the subsequent step.

[0101] If there is no suitable lighting image from the top perspective, a synthesis step is required. The lighting image from the top perspective should be an image with relatively balanced brightness and darkness. The image contains more comprehensive information, so first use the original lighting images from multiple other perspectives to average. After averaging, the brightness will be dim. From the formula below, you can see that the pixel value of the image has been weakened in the coefficient. Therefore, the coefficient is enhanced to increase the brightness and contrast of the lighting image from the top perspective. Take the illumination image including four other lighting perspectives as an example, as follows,

[0102]

[0103] Among them, I top is the illumination image from the top perspective, I TL ,I BR ,I TR ,I BL are the original illumination images under four different illumination perspectives. For example, Figure 6 The illumination images of the top view in the middle left and right images are enhanced by 0.3 times and 0.8 times respectively, that is, β is 1.3 and 1.8.

[0104] This step effectively synthesizes the top-illuminated image and enhances the coefficients based on the average value of the original image, so that the brightness and contrast enhancement are obvious.

[0105] 4) The obtained longitude image is interpolated with the synthesized top-view illumination image to obtain an illumination image of a given longitude and latitude.

[0106] From the previous step, we get the latitude as 0 and the longitude as After the longitude image is obtained, it is interpolated with the illumination image of the top view to obtain the longitude Image illuminated at a latitude of 90-θ The details are as follows:

[0107]

[0108] in The latitude is 0 and the longitude is The angle illumination image, I top is the illumination image from the top perspective, with 90-θ as the dimension.

[0109] Fig.10 The illumination image of the top perspective is enhanced by 0.8 times, the longitude is 70 degrees, and the latitudes of the left and right pictures are 20 degrees and 70 degrees respectively.

[0110] In the present invention, if there are more than 4 angles of illumination in the original illumination images, they can all be involved in the longitude image interpolation and the image difference enhancement on the same straight line.

[0111] Based on this, the present invention further proposes a multi-angle illumination image synthesis device, which is used to perform the above-mentioned multi-angle image synthesis method, comprising:

[0112] The difference module obtains several illumination images of the same sample under different illumination angles, takes the center of the illumination image as the observation center, and performs difference calculation on two illumination images whose illumination angles are on the straight line where the observation center is located to obtain several difference images. This module is used to execute the above step 1).

[0113] The first interpolation module determines the enhancement multiples according to the brightness of the original illumination images, enhances the difference image according to the determined enhancement multiples, and interpolates the obtained difference enhanced image to obtain the longitude image; this module is used to execute the above step 2).

[0114] The synthesis module calculates and processes the original illumination images to synthesize the illumination image of the top perspective; this module is used to execute the above step 3).

[0115] The second interpolation module interpolates the obtained longitude image with the synthesized top-view illumination image to obtain an illumination image of a given longitude and latitude. This module is used to execute the above step 4).

[0116] The above is only a specific implementation of the present invention, but the design concept of the present invention is not limited thereto. Any non-substantial changes to the present invention using this concept shall be deemed as an infringement of the protection scope of the present invention.

Claims

1. A multi-angle illumination image synthesis method, characterized in that: include: Acquire a plurality of illumination images of the same sample at different illumination angles, take the center of the illumination image as the observation center, and perform differential calculation on two illumination images whose illumination angles are on a straight line where the observation center is located to obtain a plurality of differential images; Determining enhancement multiples according to the brightness of the original illumination images respectively, enhancing the difference image according to the determined enhancement multiples, and performing interpolation calculation on the obtained difference enhancement image to obtain a longitude image; Acquire an illumination image from a top perspective or calculate and process a plurality of original illumination images to synthesize an illumination image from a top perspective; The obtained longitude image and the illumination image of the top viewing angle are interpolated to obtain an illumination image of given longitude and latitude.

2. A multi-angle illumination image synthesis method as claimed in claim 1, characterized in that: The pixels of the two illumination images whose illumination angles are on the straight line where the observation center is located are differentially calculated, specifically including the following: D TL =I TL -I BR ={d i,j,c ,i=1Kh,j=1Kw,c=1,2,3} d i,j,c =d i,j,c ,d i,j,c >0 d i,j,c =0,d i,j,c ≤0 Among them, I TL and I BR The illumination angle is the two illumination images of the straight line where the observation center is located, D TL is the illumination image I TL The difference image, d i,j,c is the difference image D TL The matrix elements of , i, j, c are the row number, column number and channel number of the matrix respectively, and h is the difference image D TL The height of w is the difference image D TL Width.

3. A multi-angle illumination image synthesis method as claimed in claim 1, characterized in that: The enhancement factor is determined according to the brightness of the original illumination images: if the brightness of the original illumination image is dark, the enhancement factor of the differential image corresponding to the original illumination image is increased; if the brightness of the original illumination image is bright, the enhancement factor of the differential image corresponding to the original illumination image is reduced; the brightness of the original image is calculated according to the mean of the histogram.

4. A multi-angle illumination image synthesis method as claimed in claim 1, characterized in that: The method further includes determining whether the differential enhanced image is overexposed according to the number of pixels of the differential enhanced image, and if so, adjusting the enhancement multiple of the differential image until the differential enhanced image is not overexposed, thereby obtaining the enhancement multiple.

5. The multi-angle illumination image synthesis method according to claim 1, characterized in that: The relationship between the differential enhanced image and the enhancement multiple is as follows: Among them I TL ,I BR ,I TR ,I BL is the original illumination image under four different illumination angles, α is the determined enhancement multiple, I' TL , I' TR , I' BL , I' BR are the original illumination images I under different illumination angles respectively. TL ,I BR ,I TR ,I BL The corresponding difference enhanced image.

6. The multi-angle illumination image synthesis method according to claim 1, characterized in that: The obtained differential enhanced image is interpolated to obtain the longitude image Specifically: Among them, I' TL , I' TR , I' BR , I' BL are differential enhanced images corresponding to the original illumination images under four different illumination angles, is the longitude.

7. The multi-angle illumination image synthesis method according to claim 1, characterized in that: The original illumination images are averaged and then enhanced to obtain the illumination image of the top perspective, as follows: Among them, I top is the illumination image of the top view, I TL ,I BR ,I TR ,I BL are the original illumination images under four different illumination perspectives.

8. The multi-angle illumination image synthesis method according to claim 1, characterized in that: The obtained longitude image is interpolated with the synthesized illumination image of the top perspective to obtain an illumination image of a given longitude and latitude. The details are as follows: in The latitude is 0 and the longitude is The angle illumination image, I top is the illumination image of the top viewing angle, with 90-θ as the dimension.

9. A multi-angle illumination image synthesis device, characterized in that: include: A difference module, which obtains a plurality of illumination images of the same sample at different illumination angles, takes the center of the illumination image as the observation center, and performs difference calculation on two illumination images whose illumination angles are on a straight line where the observation center is located to obtain a plurality of difference images; A first interpolation module, which determines enhancement multiples according to the brightness of the original illumination images, enhances the difference image according to the determined enhancement multiples, and performs interpolation calculation on the obtained difference enhancement image to obtain a longitude image; A synthesis module, which calculates and processes the original plurality of illumination images to synthesize an illumination image of a top perspective; The second interpolation module performs interpolation calculation on the obtained longitude image and the synthesized illumination image of the top perspective to obtain an illumination image of given longitude and latitude.

Images