An optical communication chip image enhancement method, system, device and medium based on direction-sensitive filtering and edge enhancement coupling

Abstract

The present application relates to a kind of based on direction sensitive filtering and edge enhancement coupling optical communication chip image enhancement method, system, equipment and medium, its method includes;Selective enhancement is carried out to the edge high-frequency region in the preprocessed YUV optical communication chip image by horizontal direction enhancement suppression edge enhancement coefficient, and based on the gain of dynamic adjustment enhancement mask of edge intensity, obtain spatial domain enhancement mask matrix;The difference of the multiple frequency band regions of image is sensitively enhanced, and after the gain jump of smooth low frequency and high frequency, frequency domain enhancement mask matrix is obtained;Spatial domain enhancement mask matrix and frequency domain enhancement mask matrix are fused, and directional strengthening is implemented to edge and high-frequency superposition region, and enhancement mask matrix is obtained;Brightness and color boundary constraint are carried out to superposition enhancement mask matrix, and optimal superposition enhancement mask matrix is obtained.The present application solves the problems of image blur, direction texture loss and insufficient multi-surface adaptability in optical communication chip industrial scene.

Description

Technical Field

[0001] This invention relates to the field of image enhancement technology, and in particular to an image enhancement method, system, device, and medium for optical communication chips based on direction-sensitive filtering and edge enhancement coupling. Background Technology Optical communication chip inspection is a crucial application area of ​​computer vision. Optical communication chips are typically very small, requiring high-magnification cameras for imaging. The resulting images have high resolution, complex backgrounds, and their defects often represent a small percentage of the image, posing significant challenges to subsequent image processing tasks. Limitations of industrial environments (such as uneven lighting, dust contamination, and mechanical vibration) and imaging equipment (such as line scan cameras and X-ray detectors) often result in images with low contrast, significant noise, and blurred edges, directly impacting inspection accuracy. Therefore, image enhancement methods are frequently used to improve image quality, enabling tasks such as product defect identification, dimensional measurement, and surface quality assessment.

[0002] Existing technologies mainly achieve image enhancement for optical communication chips through methods such as filtering, frequency domain enhancement, and histogram equalization, but these have significant limitations in complex industrial scenarios. (1) Insufficient directional selectivity: Traditional methods use the same enhancement strategy for horizontal and vertical edges, which cannot preserve the directional characteristics of the surface of optical devices (such as waveguide structures and electrode lines), and may introduce artifacts or blur key textures.

[0003] (2) Limitations of global parameters: Fixed parameters are difficult to adapt to the texture complexity and lighting differences of different industrial surfaces (such as the front surface, end, and back of parts). For example, using global high gain in dark areas can easily lead to noise amplification, while bright areas may be overexposed.

[0004] (3) The contradiction between computational efficiency and real-time performance: the frequency domain enhancement method requires multiple Fourier transforms, which has high computational complexity and is difficult to meet the real-time detection requirements of the pipeline (such as processing ≥10 1024×1024 images per second).

[0005] (4) Lack of preprocessing for blurred images: Existing methods do not specifically address the motion blur problem. Directly enhancing blurred images may exacerbate detail blur and reduce detection accuracy.

[0006] Therefore, those skilled in the art urgently need an image enhancement method for optical communication chips that combines direction sensitivity, adaptive adjustment capability, and efficient computation. Summary of the Invention

[0007] (a) Technical problems to be solved In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides an image enhancement method, system, device and medium for optical communication chips based on direction-sensitive filtering and edge enhancement coupling, aiming to improve the contrast, edge sharpness and detail discernibility of enhanced optical communication chip images through algorithm improvement.

[0008] (II) Technical Solution To achieve the above objectives, the main technical solutions adopted by the present invention include: In a first aspect, embodiments of the present invention provide an image enhancement method for an optical communication chip based on direction-sensitive filtering and edge enhancement coupling. The optical communication chip image includes a waveguide structure, electrode lines, and a packaging layer, comprising: The optical communication chip image is preprocessed and converted to YUV color space to obtain a YUV optical communication chip image. The edge high-frequency region in the YUV optical communication chip image is selectively enhanced by the edge enhancement coefficient of horizontal enhancement suppression, and the gain of the generated enhancement mask is dynamically adjusted based on the edge intensity to obtain the spatial enhancement mask matrix. Based on the set frequency band threshold and frequency band direction, the YUV optical communication chip image is dynamically divided into multiple frequency band regions for differentiated sensitive enhancement. The intermediate frequency transition band in the frequency band region is smoothed by the hyperbolic tangent function to smooth the gain jump between low frequency and high frequency, thus obtaining the frequency domain enhancement mask matrix. The spatial enhancement mask matrix and the frequency enhancement mask matrix are fused together. By combining the frequency distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode line, directional enhancement is performed on the edge and high-frequency superposition region. The enhancement mask matrix is ​​obtained by smoothing the gain jump between the edge and the surrounding region. By applying boundary constraints to the superimposed enhancement mask matrix based on the brightness and color constraint regions of the YUV color space determined by the optical reflection characteristics of the encapsulation layer, the optimal superimposed enhancement mask matrix for image enhancement of optical communication chips is obtained.

[0009] Optionally, the optical communication chip image is preprocessed and converted to YUV color space to obtain a YUV optical communication chip image, including: The image of the optical communication chip is enhanced by a dual sharpness enhancement process: high contrast preservation based on metal reflection suppression and USM sharpening for micron-scale structure enhancement. A brightness histogram statistical analysis was performed on the enhanced image of the optical communication chip, and a pre-contrast enhancement process was performed on the image based on the analysis results. The contrast-enhanced image of the optical communication chip is converted from the RGB color space to the YUV color space to obtain the YUV optical communication chip image.

[0010] Optionally, the edge high-frequency regions in the YUV optical communication chip image are selectively enhanced using the edge enhancement coefficient with horizontal enhancement suppression, and the gain of the generated enhancement mask is dynamically adjusted based on the edge intensity to obtain the spatial enhancement mask matrix, including: Edge detection is performed on the YUV optical communication chip image, and the edge regions in the obtained edge detection image are marked to obtain a binary mask image of the marked edge regions. The Scharr operator is used to calculate the edge gradient map of the edge detection map, and the edge intensity map and edge orientation map of the image are determined based on the edge gradient map; Set the initial edge enhancement coefficient, and adjust the enhancement coefficient of horizontal edges whose edge direction is in the horizontal edge angle range to 0.7 times the original value; Based on the adjusted edge enhancement coefficients, calculate the initial enhancement mask value for each pixel in the binary mask image; The gain of the initial enhancement mask value is dynamically adjusted based on the normalized value of the edge intensity map and the set brightness gain adjustment boundary to obtain the enhancement mask value; By outputting the enhanced mask value within the edge region identified by the binary mask image and setting the mask value in the non-edge region to a minimum, a spatial enhanced mask matrix is ​​obtained.

[0011] Optionally, based on a set frequency band threshold and frequency band direction, the YUV optical communication chip image is dynamically divided into multiple frequency band regions for differentiated sensitive enhancement. The intermediate frequency transition band within each frequency band region is smoothed using a hyperbolic tangent function to smooth the gain transition between low and high frequencies, resulting in a frequency domain enhancement mask matrix including: Based on the frequency band division benchmark set by the characteristic size of the micron-level structure of the optical communication chip, the YUV optical communication chip image is dynamically divided into low-frequency region, mid-frequency transition band and high-frequency region. Based on the frequency band direction of the high-frequency region, the high-frequency region is divided into a horizontal high-frequency region and a vertical high-frequency region; By using a directional bandpass filter bank, differentiated sensitivity enhancement is performed on the low-frequency region, mid-frequency transition band, horizontal high-frequency region, and vertical high-frequency region to obtain frequency domain enhancement images for each frequency band. Based on the frequency domain enhancement images of each frequency band, the hyperbolic tangent function is used to process the intermediate frequency transition band to smooth the gain jump between low and high frequencies, thereby obtaining the frequency domain enhancement mask matrix. The expression for the hyperbolic tangent function is: ; In the formula, R_low This is the normalized value of the distance between a pixel in the mid-frequency transition band and the boundary of the low-frequency region. R_high This is the normalized value of the distance between a pixel in the mid-frequency transition band and the boundary of the high-frequency region. smooth_low (·) smooth_high (·) represents the low-frequency gain assignment function and the high-frequency gain assignment function.

[0012] Optionally, a directional bandpass filter bank is used to perform differentiated sensitive enhancement on the low-frequency region, the mid-frequency transition band, the horizontal high-frequency region, and the vertical high-frequency region, respectively, to obtain frequency domain enhanced images for each frequency band, including: Based on the boundaries of the low-frequency region, the mid-frequency transition band, the horizontal high-frequency region, and the vertical high-frequency region, the boundaries of the images in each frequency band region are filled. Fourier transform and frequency shift are performed on the frequency band region image after image filling to obtain the frequency domain image of each frequency band; Based on the gradient direction of the frequency domain image, directional filtering of the corresponding frequency band is performed on the frequency domain image to obtain the directional filtered frequency domain image of each frequency band; Perform inverse Fourier transform on the directional filtered frequency domain image to obtain frequency domain enhanced images for each frequency band; The filled boundary regions in the frequency domain enhanced image are cropped to obtain the final frequency domain enhanced image for each frequency band; The expression for directional filtering is: ; In the formula, θ k The gradient direction value. σ 1 represents the axial bandwidth. σ 2 represents the horizontal bandwidth. H ( u , v ) represents the direction-frequency band joint selective filter function. F filtered ( u , v () represents the directional filtered frequency domain image. F ( u , v ) represents a frequency domain image.

[0013] Optionally, the spatial domain enhancement mask matrix and the frequency domain enhancement mask matrix are fused, and the edges and high-frequency superposition regions in the fused initial superimposed enhancement mask matrix are targeted to enhance the waveguide structure and the electrode line spatial orientation characteristics through the frequency domain distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode line. Finally, by smoothing the gain transition of the edges and surrounding regions, the superimposed enhancement mask matrix is ​​obtained, including: The spatial enhancement mask matrix and the frequency enhancement mask matrix are superimposed and fused. Then, based on the frequency distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode line, the edge and high-frequency superposition region in the initial superimposed enhancement mask matrix are targeted to be enhanced. The initial enhancement mask matrix is ​​processed using gradient magnitude and frequency attenuation function to smooth the gain transitions at the edges and surrounding areas, thereby obtaining the superimposed enhancement mask matrix. The expressions for the gradient magnitude and the frequency decay function are as follows: ; In the formula, R For frequency domain radius, R_high_thrd For high-frequency radius threshold, G For gradient magnitude, G 0 represents the gradient magnitude threshold.

[0014] Optionally, the optimal superimposed enhancement mask matrix for image enhancement of optical communication chips is obtained by applying boundary constraints to the superimposed enhancement mask matrix based on the brightness and color constraint regions of the YUV color space determined by the optical reflection characteristics of the encapsulation layer, including: Based on the optical reflection characteristics of the encapsulation layer, and combined with the average brightness of the Y channel in the YUV color space, a brightness constraint region is set. When the brightness of the image enhancement region is lower than that of the brightness constraint region, the gain of the superimposed enhancement mask matrix is ​​limited to 1.3 times. When the brightness of the image enhancement region is higher than that of the brightness constraint region, the gain of the superimposed enhancement mask matrix is ​​limited to 1.5 times. Obtain the area of ​​the color-connected region in the YUV optical communication chip image, and determine whether the proportion of the color-connected region area is greater than a set area proportion threshold. When the area ratio of the color-connected region exceeds the set area ratio threshold, the gain of the superimposed enhancement mask matrix will be limited to 1.1 times.

[0015] Secondly, embodiments of the present invention provide an image enhancement system for an optical communication chip based on direction-sensitive filtering and edge enhancement coupling, comprising: The image preprocessing and YUV conversion module is used to preprocess the optical communication chip image and convert it to YUV color space to obtain a YUV optical communication chip image. The spatial selective enhancement module is used to selectively enhance the high-frequency edge regions in the YUV optical communication chip image by using the edge enhancement coefficient of horizontal enhancement suppression, and to obtain the spatial enhancement mask matrix by dynamically adjusting the gain of the generated enhancement mask based on the edge intensity. The frequency domain direction-sensitive enhancement module is used to dynamically divide the YUV optical communication chip image into multiple frequency band regions for differentiated sensitive enhancement based on the set frequency band threshold and frequency band direction. The module also uses a hyperbolic tangent function to smooth the gain transition between low and high frequencies in the intermediate frequency transition band of the frequency band region to obtain a frequency domain enhancement mask matrix. The mask fusion module is used to fuse the spatial domain enhancement mask matrix and the frequency domain enhancement mask matrix. By using the frequency domain distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode lines, it performs directional enhancement on the edges and high-frequency superposition regions in the initial superimposed enhancement mask matrix obtained by fusion. Finally, by smoothing the gain transition of the edges and surrounding regions, the superimposed enhancement mask matrix is ​​obtained. The boundary constraint optimization module is used to perform boundary constraints on the superimposed enhancement mask matrix based on the brightness and color constraint regions of the YUV color space determined by the optical reflection characteristics of the encapsulation layer, so as to obtain the optimal superimposed enhancement mask matrix for image enhancement of optical communication chips.

[0016] Thirdly, embodiments of the present invention provide an electronic device, comprising: processor; The memory stores the steps of the optical communication chip image enhancement method based on direction-sensitive filtering and edge enhancement coupling described above, which are used by the processor to control the process.

[0017] Fourthly, embodiments of the present invention provide a computer-readable medium storing computer-executable instructions thereon, which, when executed by a processor, implement the steps of the optical communication chip image enhancement method based on direction-sensitive filtering and edge enhancement coupling described above.

[0018] (III) Beneficial Effects The beneficial effects of this invention are as follows: By employing a selective enhancement strategy for the edge high-frequency region with horizontal enhancement suppression, this invention separates the horizontal / vertical high-frequency gain, effectively preserving the directional texture of the optical communication chip surface, avoiding the detail blurring caused by traditional isotropic filtering, and combining it with dynamic adjustment of gain based on edge intensity, it better distinguishes noise and effective high-frequency characteristics, reducing the over-enhancement phenomenon in high-noise regions.

[0019] Meanwhile, the present invention also adopts a multi-band dynamic division mechanism to achieve adaptive multi-scene adaptation. The threshold is dynamically calculated to divide the frequency band according to the surface type of the detected image, and the mid-frequency transition band in the frequency band region is smoothed by the hyperbolic tangent function to smooth the gain jump between low frequency and high frequency, so as to retain the directional texture of the optical communication chip surface through differential gain, thus breaking through the limitations of traditional isotropic filtering. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating an image enhancement method for an optical communication chip based on direction-sensitive filtering and edge enhancement coupling, according to an embodiment of the present invention. Figure 2 A flowchart illustrating an image enhancement method for an optical communication chip based on direction-sensitive filtering and edge enhancement coupling, provided as an embodiment of the present invention; Figure 3A visual comparison image of an optical communication chip provided in an embodiment of the present invention before and after preprocessing; Figure 4 This is a binary mask image for identifying the edge region of an optical communication chip, provided in an embodiment of the present invention. Figure 5 A visual comparison image of an optical communication chip before and after frequency domain differential sensitivity enhancement provided in an embodiment of the present invention; Figures 6 to 9 This is a visual comparison image of different types of optical communication chip images before and after image enhancement processing, provided in an embodiment of the present invention. Detailed Implementation

[0021] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] refer to Figures 1 to 9 As shown in the embodiment of the present invention, an image enhancement method for optical communication chips based on direction-sensitive filtering and edge enhancement coupling is proposed. The optical communication chip image includes a waveguide structure, electrode lines, and a packaging layer. The method includes: preprocessing the optical communication chip image and converting it to YUV color space to obtain a YUV optical communication chip image; selectively enhancing the high-frequency edge regions in the YUV optical communication chip image using an edge enhancement coefficient that suppresses horizontal direction enhancement, and dynamically adjusting the gain of the generated enhancement mask based on the edge intensity to obtain a spatial enhancement mask matrix; dynamically dividing the YUV optical communication chip image into multiple frequency band regions for differentiated sensitive enhancement according to a set frequency band threshold and frequency band direction, and... The intermediate frequency transition band in the frequency band region is smoothed by using a hyperbolic tangent function to smooth the gain transition between low and high frequencies, thus obtaining a frequency domain enhancement mask matrix. The spatial domain enhancement mask matrix and the frequency domain enhancement mask matrix are fused, and the edges and high-frequency superposition regions in the fused initial superimposed enhancement mask matrix are targeted for enhancement based on the frequency domain distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode lines. Finally, the superimposed enhancement mask matrix is ​​obtained by smoothing the gain transition between the edges and the surrounding regions. The superimposed enhancement mask matrix is ​​then subjected to boundary constraints based on the brightness and color constraint regions of the YUV color space determined by the optical reflection characteristics of the encapsulation layer, thus obtaining the optimal superimposed enhancement mask matrix for image enhancement of optical communication chips.

[0023] This embodiment employs a selective enhancement strategy for edge high-frequency regions with horizontal enhancement suppression, separating horizontal and vertical high-frequency gains. This effectively preserves the directional texture of the optical communication chip surface, avoiding detail blurring caused by traditional isotropic filtering. Furthermore, by dynamically adjusting the gain based on edge intensity, it better distinguishes noise and effective high-frequency characteristics, reducing over-enhancement in high-noise regions.

[0024] Meanwhile, this embodiment also adopts a multi-band dynamic division mechanism to achieve adaptive multi-scene adaptation. The threshold is dynamically calculated and the frequency band is divided according to the surface type of the detected image. The intermediate frequency transition band in the frequency band region is smoothed by the hyperbolic tangent function to smooth the gain jump between low frequency and high frequency, so as to preserve the directional texture of the optical communication chip surface through differentiated gain, thus breaking through the limitations of traditional isotropic filtering.

[0025] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0026] Specifically, refer to Figure 1 As shown, this embodiment provides an image enhancement method for optical communication chips based on direction-sensitive filtering and edge enhancement coupling. The optical communication chip image includes a waveguide structure, electrode lines, and a packaging layer. The method includes: S100. Preprocess the optical communication chip image and convert it to YUV color space to obtain a YUV optical communication chip image.

[0027] In this embodiment, step S100 may include the following sub-steps S110 to S130: S110 performs dual sharpness enhancement processing on the image of the optical communication chip by high contrast preservation based on metal reflection suppression and USM sharpening for micron-scale structure enhancement.

[0028] To address the need for improved clarity of micron-level structures and interference from metallic reflections in optical communication chip imaging, this implementation employs a two-stage enhancement strategy. First, based on high-contrast preservation technology that suppresses metallic reflections, it intelligently identifies and suppresses overexposure artifacts and halo effects in strongly reflective areas of the chip surface (such as electrode lines and pads) generated by traditional high-contrast filtering, producing a substrate image that preserves the true edges of the structure. Then, it uses Unsharp Mask (USM) sharpening for micron-level structure enhancement, utilizing a small-radius Gaussian kernel and adaptive threshold parameters to precisely improve the edge contrast of micron-level features (such as waveguides, grating lines, and electrode gaps) while suppressing noise interference. This dual synergistic effect significantly overcomes the technical challenges of reflection interference and blurring of fine structures.

[0029] S120. Perform brightness histogram statistical analysis on the enhanced optical communication chip image, and perform pre-contrast enhancement processing on the optical communication chip image based on the analysis results.

[0030] For images that are generally dark or blurry, statistical analysis of the brightness histogram can be used to determine if pre-contrast enhancement is necessary. For example, the brightness can be divided into eight segments, and their brightness distribution histograms can be analyzed. If any segment accounts for more than 60% of the total, or the difference between the highest and lowest percentages is less than 10%, then pre-contrast enhancement, i.e., adaptive histogram equalization, is performed. Figure 3 As shown, this is a visual comparison of an optical communication chip image before and after preprocessing. The left side is the original image before preprocessing, and the right side is the image after processing.

[0031] S130. The contrast-enhanced optical communication chip image is converted from the RGB color space to the YUV color space to obtain the YUV optical communication chip image.

[0032] The expression for image color space conversion is: (1) S200: Selectively enhance the high-frequency edge regions in the YUV optical communication chip image by using the horizontal enhancement suppression edge enhancement coefficient, and obtain the spatial enhancement mask matrix by dynamically adjusting the gain of the generated enhancement mask based on the edge intensity.

[0033] This embodiment employs a selective enhancement strategy for edge high-frequency regions with horizontal enhancement suppression, separating horizontal and vertical high-frequency gains to effectively preserve the directional texture of the optical communication chip surface, avoiding detail blurring caused by traditional isotropic filtering. Furthermore, by dynamically adjusting the gain based on edge intensity, it better distinguishes noise and effective high-frequency characteristics, reducing over-enhancement in high-noise regions.

[0034] In this embodiment, step S200 may include the following sub-steps S210 to S260: S210. Perform edge detection on the YUV optical communication chip image, and mark the edge regions in the obtained edge detection map to obtain a binary mask map of the marked edge regions.

[0035] For example, using the Canny operator pair Figure 3 The image of the optical communication chip shown is subjected to edge detection. Then, the detected edge regions and non-edge regions are masked to obtain the image as shown. Figure 4 The binary mask shown.

[0036] S220. Calculate the edge gradient map of the edge detection map using the Scharr operator, and determine the edge intensity map and edge orientation map of the image based on the edge gradient map.

[0037] Furthermore, the Scharr operator is first used to calculate the horizontal and vertical gradients in the edge detection map. Next, the Euclidean distance is calculated for the gradient components of each pixel to obtain the edge intensity map. Then, the gradient direction of each pixel is calculated and transformed by angle calculation to obtain the edge direction map, with a calculation range of 0~2π.

[0038] S230. Set the initial edge enhancement coefficient, and adjust the enhancement coefficient of the horizontal edge in the horizontal edge angle range to 0.7 times the original value.

[0039] Furthermore, for horizontal edges with an edge direction of ±20 degrees, the enhancement coefficient of the horizontal edge is reduced by 30%, that is, the enhancement coefficient of the horizontal edge is adjusted to 0.7 times the original value, thereby avoiding horizontal stripe enhancement.

[0040] S240. Based on the adjusted edge enhancement coefficient, calculate the initial enhancement mask value for each pixel in the binary mask image.

[0041] S250. The gain of the initial enhancement mask value is dynamically adjusted based on the normalized value of the edge intensity map and the set brightness gain adjustment boundary to obtain the enhancement mask value.

[0042] Furthermore, the edge intensity map is normalized to 0-1. The gain of the initial enhancement mask value is dynamically adjusted by combining the generated normalized value with a boundary limiting the dark area gain to 1.3, thereby avoiding noise amplification. Specifically, regions in the YUV optical communication chip image with an average brightness of less than 32 in the Y channel are defined as dark areas.

[0043] S260. Output the enhanced mask value in the edge region identified by the binary mask image and set the mask value in the non-edge region to a minimum value to obtain the spatial enhanced mask matrix.

[0044] S300: Based on the set frequency band threshold and frequency band direction, the YUV optical communication chip image is dynamically divided into multiple frequency band regions for differentiated sensitive enhancement. The intermediate frequency transition band in the frequency band region is smoothed by the hyperbolic tangent function to smooth the gain jump between low frequency and high frequency, thereby obtaining the frequency domain enhancement mask matrix.

[0045] This embodiment employs a dynamic frequency band partitioning mechanism to achieve adaptive multi-scene adaptation. It dynamically calculates thresholds to partition frequency bands based on the surface type of the detected image, and smooths the gain transition between low and high frequencies using a hyperbolic tangent function in the mid-frequency transition band within the frequency band region. This differentiated gain preserves the directional texture of the optical communication chip surface, overcoming the limitations of traditional isotropic filtering. Figure 5As shown, the left side is the original spectrogram, and the right side is the enhanced spectrogram. It can be seen that the enhanced spectrogram can faintly reveal traces of characters and edge contours. This indicates that the differentially sensitive enhancement operation has a significant impact on the frequency components of the image, making the character-related frequency information that was not obvious in the original spectrogram more prominent.

[0046] In this embodiment, step S300 may include the following sub-steps S310 to S340: S310: Based on the frequency band division benchmark set by the characteristic size of the micron-level structure of the optical communication chip, the YUV optical communication chip image is dynamically divided into low-frequency region, mid-frequency transition band and high-frequency region.

[0047] Furthermore, firstly, an initial value for frequency band division is selected based on a certain proportion of the frequency domain distance matrix. Then, the image resolution, average brightness, and image complexity are adaptively adjusted according to different weights to obtain the frequency band division benchmark set by the feature size of the micron-level structure of the optical communication chip. The low-frequency region is the low-frequency component in the image signal, that is, the region where the image intensity (brightness, grayscale) changes smoothly; the high-frequency region is the high-frequency component in the image signal, that is, the region where the image signal intensity (brightness / grayscale) changes drastically; the mid-frequency transition band is the region set between the low-frequency and high-frequency regions.

[0048] S320. Based on the frequency band direction of the high-frequency region, the high-frequency region is divided into a horizontal high-frequency region and a vertical high-frequency region.

[0049] Furthermore, since the human eye is more sensitive to the horizontal and vertical directions, and the human eye is usually more sensitive to the horizontal direction, the high-frequency band, which usually has more object edges and details, is processed differently in the horizontal and vertical directions to further reduce the limitations of directional filtering.

[0050] S330 utilizes a directional bandpass filter bank to perform differentiated sensitivity enhancement on the low-frequency region, mid-frequency transition band, horizontal high-frequency region, and vertical high-frequency region, respectively, to obtain frequency domain enhancement images for each frequency band.

[0051] Further, step S330 may include the following sub-steps S331 to S335: S331. Based on the boundaries of the low-frequency region, the mid-frequency transition band, the horizontal high-frequency region, and the vertical high-frequency region, fill the boundaries of the image of each frequency band region.

[0052] To avoid aliasing, the image is resized to its original size. HxW Fill the dimensions with powers of 2, specifically the powers. m The image fill expression is set according to the actual image resolution: (2) In equation (2), I padded Represents filling an image with another image. I Represents the original image. ZeroPadding (·) represents the image fill calculation function. size Represents dimensions.

[0053] S332. Perform Fourier transform and frequency shift on the frequency band region image after image filling to obtain the frequency domain image of each frequency band.

[0054] The image is transformed from the spatial domain to the frequency domain using the Fourier transform function, and then the zero-frequency component after the Fourier transform is shifted to the center of the spectrum using the Fourier zero-frequency shift function. (3) In equation (3), F ( u , v ) represents the frequency domain image. FFTShift (·) represents the Fourier zero-frequency shift function. FFT (·) represents the Fourier transform function.

[0055] S333. Based on the gradient direction of the frequency domain image, perform directional filtering on the frequency domain image for the corresponding frequency band to obtain the directional filtered frequency domain image for each frequency band.

[0056] The expression for directional filtering is: (4) In equation (4), θ k The gradient direction is quantized into four main directions: horizontal (0 / 180°), vertical (90 / 270°), left diagonal (45 / 235°), and right diagonal (135 / 315°). σ 1 represents the axial bandwidth, the width of the principal axis along the gradient direction; σ 2 represents the horizontal bandwidth, and the secondary axis width in the vertical gradient direction; H ( u , v ) represents the direction-band joint selective filter function; F filtered ( u , v () represents the directional filtered frequency domain image. S334. Perform inverse Fourier transform on the directional filtered frequency domain image to obtain the frequency domain enhanced image for each frequency band.

[0057] Inverse Fourier Transform: (5) In equation (5),F’ ( u , v () represents a frequency domain enhanced image. IFFT (·) represents the inverse Fourier transform function. FFTShift -1 (·) represents the Fourier inverse zero-frequency shift function.

[0058] S335. Crop the filled boundary regions in the frequency domain enhanced image to obtain the final frequency domain enhanced image for each frequency band.

[0059] (6) In equation (6), I enhanced This represents the cropped image. Crop (·) represents the image cropping function, used to crop and fill boundary regions.

[0060] S340. Based on the frequency domain enhancement images of each frequency band, the intermediate frequency transition band is processed using the hyperbolic tangent function to smooth the gain jump between low and high frequencies, thereby obtaining the frequency domain enhancement mask matrix.

[0061] The expression for the hyperbolic tangent function is: (7) In equation (7), R_low This is the normalized value of the distance between a pixel in the mid-frequency transition band and the boundary of the low-frequency region. R_high This is the normalized value of the distance between a pixel in the mid-frequency transition band and the boundary of the high-frequency region. smooth_low (·) smooth_high (·) represents the low-frequency gain assignment function and the high-frequency gain assignment function.

[0062] S400. The spatial domain enhancement mask matrix and the frequency domain enhancement mask matrix are fused. The edge and high-frequency superposition region in the initial superposition enhancement mask matrix are directionally enhanced by the frequency domain distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode line. Finally, the superposition enhancement mask matrix is ​​obtained by smoothing the gain jump of the edge and the surrounding region.

[0063] In this embodiment, the spatial domain enhancement mask matrix and the frequency domain enhancement mask matrix are superimposed and fused to enhance the frequency domain distribution feature region (high frequency region) belonging to the waveguide structure and the spatial orientation feature region (edge ​​region) belonging to the electrode line, thereby further highlighting the high frequency edge region of the waveguide structure and the electrode line.

[0064] In this embodiment, step S400 may include the following sub-steps S410 to S420: S410. The spatial enhancement mask matrix and the frequency enhancement mask matrix are superimposed and fused. Based on the frequency distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode line, directional enhancement is performed on the edge and high-frequency superposition region in the initial superimposed enhancement mask matrix obtained by superposition and fusion.

[0065] S420. Using the gradient magnitude and frequency attenuation function, the initial enhancement mask matrix is ​​processed to smooth the gain transitions at the edges and surrounding areas, thereby obtaining the superimposed enhancement mask matrix.

[0066] The expressions for the gradient magnitude and frequency decay function are as follows: (8) In equation (8), R For frequency domain radius, R_high_thrd For high-frequency radius threshold, G For gradient magnitude, G 0 represents the gradient magnitude threshold.

[0067] S500: Based on the brightness and color constraint regions of the YUV color space determined by the optical reflection characteristics of the encapsulation layer, the superimposed enhancement mask matrix is ​​subjected to boundary constraints to obtain the optimal superimposed enhancement mask matrix for image enhancement of optical communication chips.

[0068] In order to avoid excessive brightness or color enhancement that could lead to color distortion, this embodiment requires some restrictions on the mask matrix. It employs detection of excessively dark or bright areas and large areas of solid color, and limits the gain of these areas based on the detection results.

[0069] In this embodiment, step S500 may include the following sub-steps S510 to S540: S510. Based on the optical reflection characteristics of the encapsulation layer, and combined with the average brightness of the Y channel in the YUV color space, a brightness constraint area is set.

[0070] Furthermore, firstly, an initial brightness constraint region is set based on the average brightness of the Y channel in the YUV color space (less than 32 is too dark, and greater than 230 is too bright). Then, the initial brightness constraint region is adaptively adjusted through the optical reflection characteristics of the encapsulation layer.

[0071] S520a. When the brightness of the image enhancement region is lower than that of the brightness constraint region, the gain of the superimposed enhancement mask matrix is ​​limited to 1.3 times.

[0072] S520b: When the brightness of the image enhancement region is higher than that of the brightness constraint region, the gain of the superimposed enhancement mask matrix is ​​limited to 1.5 times.

[0073] S530: Obtain the area of ​​the color-connected region in the YUV optical communication chip image, and determine whether the proportion of the color-connected region area is greater than the set area proportion threshold.

[0074] Furthermore, large-area solid color region detection is adopted. Based on whether the area of ​​the color connected region is greater than 10% of the effective area of ​​the whole image, the connected region greater than 10% of the effective area is identified as a solid color region, and the gain of the solid color region is limited to 1.1. The main colors identified are yellow, blue, purple and green.

[0075] S540. When the area ratio of the color connected region is greater than the set area ratio threshold, the gain of the superimposed enhancement mask matrix is ​​limited to 1.1 times.

[0076] In one specific embodiment, image enhancement processing steps S100 to S500 are performed on images of four types of optical communication chips. A visual comparison image before and after the enhancement processing is provided below. Figures 6 to 9 As shown in Table 1, the left side is the original blurred image, and the right side is the image enhancement result after the method of this embodiment. The image enhancement effect is evaluated using both technical indicators and visual effects, as shown in Table 1. Figure 5 A comparison of PSNR and SSIM metrics for image enhancement processing of four types of optical communication chips. PSNR (Peak Signal-to-Noise Ratio) and SSIM (Structural Similarity) are two commonly used image quality evaluation metrics. PSNR is calculated based on mean squared error (MSE); a higher PSNR value indicates better image quality, meaning smaller differences between images. However, PSNR does not accurately reflect human perception of image quality because it is based solely on pixel-level error calculations. SSIM, on the other hand, considers the human eye's greater sensitivity to structural information in images. Therefore, it evaluates image similarity from three aspects: brightness, contrast, and structure. An SSIM value closer to 1 indicates greater image similarity.

[0077] Table 1. Comparison of PSNR and SSIM indices for image enhancement processing of different types of optical communication chips.

[0078] Combination Figure 6 , 7 Compared with the analysis in Table 1, for Figure 6 Left side and Figure 7In terms of technical specifications, the two types of images on the right have PSNR in the middle range of 25-30dB, which may indicate some pixel distortion or compression artifacts; however, their SSIM is close to 0.9, indicating that the structural information of the images is well preserved. From a visual perspective, these two types of images have relatively complex textures, resulting in improved clarity and depth. Figure 8 Left side and Figure 9 It's more obvious on the right; for Figure 8 Left side and Figure 9 In terms of technical indicators, the two types of images on the right have significantly higher PSNR values ​​than the conventional threshold (above 30dB is considered good), indicating that pixel-level distortion is almost negligible; their average SSIM is above 0.9, indicating that the image's brightness, contrast, and structural information are well preserved, and the visual effect is close to the original image. However, in terms of visual effect, due to the relatively simple image structure, the overall improvement is not as significant as... Figure 6 Left side and Figure 7 The two types of images on the right are more intuitive.

[0079] Furthermore, this embodiment also provides an image enhancement system for an optical communication chip based on direction-sensitive filtering and edge enhancement coupling, comprising: The image preprocessing and YUV conversion module is used to preprocess the optical communication chip image and convert it to YUV color space to obtain a YUV optical communication chip image. The spatial selective enhancement module is used to selectively enhance the high-frequency edge regions in the YUV optical communication chip image by using the edge enhancement coefficient of horizontal enhancement suppression, and to obtain the spatial enhancement mask matrix by dynamically adjusting the gain of the generated enhancement mask based on the edge intensity. The frequency domain direction-sensitive enhancement module is used to dynamically divide the YUV optical communication chip image into multiple frequency band regions for differentiated sensitive enhancement based on the set frequency band threshold and frequency band direction. The module also uses a hyperbolic tangent function to smooth the gain transition between low and high frequencies in the intermediate frequency transition band of the frequency band region to obtain a frequency domain enhancement mask matrix. The mask fusion module is used to fuse the spatial domain enhancement mask matrix and the frequency domain enhancement mask matrix. By using the frequency domain distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode lines, it performs directional enhancement on the edges and high-frequency superposition regions in the initial superimposed enhancement mask matrix obtained by fusion. Finally, by smoothing the gain transition of the edges and surrounding regions, the superimposed enhancement mask matrix is ​​obtained. The boundary constraint optimization module is used to perform boundary constraints on the superimposed enhancement mask matrix based on the brightness and color constraint regions of the YUV color space determined by the optical reflection characteristics of the encapsulation layer, so as to obtain the optimal superimposed enhancement mask matrix for image enhancement of optical communication chips.

[0080] Furthermore, this embodiment also provides an electronic device, including: a processor; and a memory storing steps for the processor to control the above-described optical communication chip image enhancement method based on direction-sensitive filtering and edge enhancement coupling.

[0081] Finally, this embodiment also provides a computer-readable medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the optical communication chip image enhancement method based on direction-sensitive filtering and edge enhancement coupling described above.

[0082] In summary, this embodiment provides an image enhancement method, system, device, and medium for optical communication chips based on direction-sensitive filtering and edge enhancement coupling. For optical communication chip images with different brightness distributions and varying degrees of complexity in complex industrial scenarios, it first enhances blurred image preprocessing, then uses direction-sensitive filtering to preserve low-frequency background, enhance high-frequency directional edges, and smooth mid-frequency transitions in the industrial image. Combined with adaptive parameter configuration, it finds the appropriate strategy to achieve image enhancement, solving the problems of image blurring, loss of directional texture, and insufficient adaptability to multiple surfaces in industrial scenarios for optical communication chips. It improves the edge clarity and detail detectability of industrial images, and is further applicable to real-time image preprocessing in automated inspection lines and better suited for downstream tasks such as defect detection.

[0083] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0084] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0085] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0086] It should be noted that in the description of this invention, the word "a" or "an" preceding a component does not exclude the existence of multiple such components. This invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. The use of terms such as first, second, third, etc., is merely for convenience and does not indicate any order. These terms can be understood as part of the component names.

[0087] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0088] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning of the basic inventive concept, can make other changes and modifications to these embodiments.

[0089] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from the spirit and scope of the invention.

Claims

1. A method for enhancing the image of an optical communication chip based on direction-sensitive filtering and edge enhancement coupling, wherein the optical communication chip image includes a waveguide structure, electrode lines, and a packaging layer, characterized in that, include: The optical communication chip image is preprocessed and converted to YUV color space to obtain a YUV optical communication chip image. This method selectively enhances high-frequency edge regions in a YUV optical communication chip image using a horizontally enhanced edge enhancement coefficient. The gain of the generated enhancement mask is dynamically adjusted based on the edge intensity to obtain a spatial enhancement mask matrix. The process includes: edge detection of the YUV optical communication chip image, marking edge regions in the resulting edge detection map to obtain a binary mask map of the marked edge regions; calculating the edge gradient map of the edge detection map using the Scharr operator, and determining the edge intensity map and edge direction map of the image based on the edge gradient map; setting initial edge enhancement coefficients, and adjusting the enhancement coefficients of horizontal edges with edge directions in the horizontal edge angle range to 0.7 times the original value; calculating the initial enhancement mask value for each pixel in the binary mask map based on the adjusted edge enhancement coefficients; dynamically adjusting the gain of the initial enhancement mask value according to the normalized value of the edge intensity map and the set brightness gain adjustment boundary to obtain the enhancement mask value; outputting the enhancement mask value within the edge regions marked in the binary mask map and setting the mask value in non-edge regions to a minimum value to obtain a spatial enhancement mask matrix. Based on the characteristic dimensions of the micron-level structure of the optical communication chip, the frequency band threshold and frequency band direction are set, and the YUV optical communication chip image is dynamically divided into multiple frequency band regions for differential sensitive enhancement. The intermediate frequency transition band in the frequency band region is smoothed by the hyperbolic tangent function to smooth the gain jump between low frequency and high frequency, and the frequency domain enhancement mask matrix is ​​obtained. The spatial domain enhancement mask matrix and the frequency domain enhancement mask matrix are fused. Then, based on the frequency domain distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode line, the edges and high-frequency superposition regions in the initial superimposed enhancement mask matrix are targeted to be enhanced. Finally, the superimposed enhancement mask matrix is ​​obtained by smoothing the gain jump of the edges and surrounding regions. Based on the brightness and color constraint regions of the YUV color space determined by the optical reflection characteristics of the encapsulation layer, the superimposed enhancement mask matrix is ​​subjected to boundary constraints to obtain the optimal superimposed enhancement mask matrix for image enhancement of optical communication chips. This includes: setting a brightness constraint region based on the optical reflection characteristics of the encapsulation layer and the average brightness of the Y channel in the YUV color space; limiting the gain of the superimposed enhancement mask matrix to 1.3 times when the brightness of the image enhancement region is lower than that of the brightness constraint region; limiting the gain of the superimposed enhancement mask matrix to 1.5 times when the brightness of the image enhancement region is higher than that of the brightness constraint region; obtaining the area of ​​the color connected region in the YUV optical communication chip image and determining whether the proportion of the color connected region area is greater than a set area proportion threshold; limiting the gain of the superimposed enhancement mask matrix to 1.1 times when the proportion of the color connected region area is greater than the set area proportion threshold.

2. The method as described in claim 1, characterized in that, Preprocessing and YUV color space conversion of the optical communication chip image to obtain a YUV optical communication chip image includes: The image of the optical communication chip is enhanced by a dual sharpness enhancement process: high contrast preservation based on metal reflection suppression and USM sharpening for micron-scale structure enhancement. A brightness histogram statistical analysis was performed on the enhanced image of the optical communication chip, and a pre-contrast enhancement process was performed on the image based on the analysis results. The contrast-enhanced image of the optical communication chip is converted from the RGB color space to the YUV color space to obtain the YUV optical communication chip image.

3. The method as described in claim 1, characterized in that, Based on the characteristic dimensions of the micron-level structure of the optical communication chip, the frequency band threshold and frequency band direction are set to dynamically divide the YUV optical communication chip image into multiple frequency band regions for differential sensitive enhancement. The intermediate frequency transition band within each frequency band region is smoothed using a hyperbolic tangent function to smooth the gain transition between low and high frequencies, resulting in a frequency domain enhancement mask matrix including: Based on the frequency band division benchmark set by the characteristic size of the micron-level structure of the optical communication chip, the YUV optical communication chip image is dynamically divided into low-frequency region, mid-frequency transition band and high-frequency region. Based on the frequency band direction of the high-frequency region, the high-frequency region is divided into a horizontal high-frequency region and a vertical high-frequency region; By using a directional bandpass filter bank, differentiated sensitivity enhancement is performed on the low-frequency region, mid-frequency transition band, horizontal high-frequency region, and vertical high-frequency region to obtain frequency domain enhancement images for each frequency band. Based on the frequency domain enhancement images of each frequency band, the hyperbolic tangent function is used to process the intermediate frequency transition band to smooth the gain jump between low and high frequencies, thereby obtaining the frequency domain enhancement mask matrix. The expression for the hyperbolic tangent function is: ； In the formula, R_low This is the normalized value of the distance between a pixel in the mid-frequency transition band and the boundary of the low-frequency region. R_high This is the normalized value of the distance between a pixel in the mid-frequency transition band and the boundary of the high-frequency region. smooth_low (·) smooth_high (·) represents the low-frequency gain assignment function and the high-frequency gain assignment function.

4. The method as described in claim 3, characterized in that, Differential sensitivity enhancement is performed on the low-frequency region, mid-frequency transition band, horizontal high-frequency region, and vertical high-frequency region using a directional bandpass filter bank to obtain frequency domain enhanced images for each frequency band, including: Based on the boundaries of the low-frequency region, the mid-frequency transition band, the horizontal high-frequency region, and the vertical high-frequency region, the boundaries of the images in each frequency band region are filled. Fourier transform and frequency shift are performed on the frequency band region image after image filling to obtain the frequency domain image of each frequency band; Based on the gradient direction of the frequency domain image, directional filtering of the corresponding frequency band is performed on the frequency domain image to obtain the directional filtered frequency domain image of each frequency band; Perform inverse Fourier transform on the directional filtered frequency domain image to obtain frequency domain enhanced images for each frequency band; The filled boundary regions in the frequency domain enhanced image are cropped to obtain the final frequency domain enhanced image for each frequency band; The expression for directional filtering is: ； In the formula, θ k The gradient direction value. σ 1 represents the axial bandwidth. σ 2 represents the horizontal bandwidth. H ( u , v ) represents the direction-frequency band joint selective filter function. F filtered ( u , v () represents the directional filtered frequency domain image. F ( u , v ) represents a frequency domain image.

5. The method as described in claim 1, characterized in that, The spatial enhancement mask matrix and the frequency enhancement mask matrix are fused. Then, based on the frequency distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode lines, directional enhancement is applied to the edges and high-frequency superposition regions in the fused initial superimposed enhancement mask matrix. Finally, by smoothing the gain transitions of the edges and surrounding regions, the superimposed enhancement mask matrix is ​​obtained, including: The spatial enhancement mask matrix and the frequency enhancement mask matrix are superimposed and fused. Then, based on the frequency distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode line, the edge and high-frequency superposition region in the initial superimposed enhancement mask matrix are targeted to be enhanced. The initial enhancement mask matrix is ​​processed using gradient magnitude and frequency attenuation function to smooth the gain transitions at the edges and surrounding areas, thereby obtaining the superimposed enhancement mask matrix. The expressions for the gradient magnitude and the frequency decay function are as follows: ； In the formula, R For frequency domain radius, R_high_thrd For high-frequency radius threshold, G For gradient magnitude, G 0 represents the gradient magnitude threshold.

6. An image enhancement system for an optical communication chip based on direction-sensitive filtering and edge enhancement coupling, characterized in that, include: The image preprocessing and YUV conversion module is used to preprocess the optical communication chip image and convert it to YUV color space to obtain a YUV optical communication chip image. A spatially selective enhancement module is used to selectively enhance high-frequency edge regions in a YUV optical communication chip image by using horizontally enhanced and suppressed edge enhancement coefficients. It dynamically adjusts the gain of the generated enhancement mask based on edge intensity to obtain a spatial enhancement mask matrix. The module includes: edge detection of the YUV optical communication chip image, marking edge regions in the obtained edge detection map to obtain a binary mask map of the marked edge regions; calculating the edge gradient map of the edge detection map using the Scharr operator, and determining the edge intensity map and edge direction map of the image based on the edge gradient map; setting initial edge enhancement coefficients, and adjusting the enhancement coefficients of horizontal edges with edge directions in the horizontal edge angle range to 0.7 times the original value; calculating the initial enhancement mask value for each pixel in the binary mask map based on the adjusted edge enhancement coefficients; dynamically adjusting the gain of the initial enhancement mask value according to the normalized value of the edge intensity map and the set brightness gain adjustment boundary to obtain the enhancement mask value; outputting the enhancement mask value within the edge regions marked in the binary mask map and setting the mask value in non-edge regions to a minimum value to obtain a spatial enhancement mask matrix. The frequency domain orientation sensitive enhancement module is used to dynamically divide the YUV optical communication chip image into multiple frequency band regions for differentiated sensitive enhancement based on the frequency band threshold and frequency band orientation set according to the characteristic size of the micron-level structure of the optical communication chip. The module also smooths the gain transition between low and high frequencies in the mid-frequency transition band of the frequency band region through the hyperbolic tangent function to obtain the frequency domain enhancement mask matrix. The mask fusion module is used to fuse the spatial domain enhancement mask matrix and the frequency domain enhancement mask matrix. By using the frequency domain distribution characteristics of the waveguide structure and the spatial orientation characteristics of the electrode lines, it performs directional enhancement on the edges and high-frequency superposition regions in the initial superimposed enhancement mask matrix obtained by fusion. Finally, by smoothing the gain transition of the edges and surrounding regions, the superimposed enhancement mask matrix is ​​obtained. The boundary constraint optimization module is used to apply boundary constraints to the superimposed enhancement mask matrix based on the brightness and color constraint regions of the YUV color space determined by the optical reflection characteristics of the encapsulation layer, thereby obtaining the optimal superimposed enhancement mask matrix for image enhancement of the optical communication chip. This includes: setting a brightness constraint region based on the optical reflection characteristics of the encapsulation layer and the average brightness of the Y channel in the YUV color space; limiting the gain of the superimposed enhancement mask matrix to 1.3 times when the brightness of the image enhancement region is lower than that of the brightness constraint region; limiting the gain of the superimposed enhancement mask matrix to 1.5 times when the brightness of the image enhancement region is higher than that of the brightness constraint region; obtaining the area of ​​the color-connected region in the YUV optical communication chip image and determining whether the proportion of the color-connected region area is greater than a set area proportion threshold; and limiting the gain of the superimposed enhancement mask matrix to 1.1 times when the proportion of the color-connected region area is greater than the set area proportion threshold.

7. An electronic device, characterized in that, include: processor; The memory stores the steps of the optical communication chip image enhancement method based on direction-sensitive filtering and edge enhancement coupling as described in any one of claims 1-5, for processor control.

8. A computer-readable medium having computer-executable instructions stored thereon, characterized in that, When the executable instructions are executed by the processor, they implement the steps of the optical communication chip image enhancement method based on direction-sensitive filtering and edge enhancement coupling as described in any one of claims 1-5.

Images