Image quality improvement method and system based on liquid crystal microlens array

Abstract

The application discloses an image quality improving method and system based on a liquid crystal microlens array, and comprises the following steps: forming initial image data; performing microlens response gridding processing to generate an image quality parameter set; performing quality deviation identification by using an improved iDETEX model to generate a quality deviation data set; generating array corresponding relationship data; determining the focal length correction amount, phase correction amount and convergence correction amount corresponding to each liquid crystal microlens unit to generate a liquid crystal microlens array regulation parameter set; constructing a liquid crystal microlens array driving matrix; generating regulated image data; and performing focal plane detail fusion processing to generate a target high-quality image. The application improves the local regulation precision and image quality of liquid crystal microlens array imaging.

Description

Technical Field

[0001] This invention relates to the field of liquid crystal optical imaging and image processing, and in particular to a method and system for improving image quality based on a liquid crystal microlens array. Background Technology

[0002] With the widespread application of optical imaging equipment in machine vision, intelligent inspection, medical imaging, miniature camera modules, light field imaging, and precision measurement, the requirements for image sharpness, edge preservation, brightness uniformity, local detail integrity, and geometric imaging stability are constantly increasing. Traditional imaging systems typically rely on fixed lens groups, mechanical focusing structures, or back-end image processing algorithms to improve image quality. Fixed lens groups mainly achieve imaging through a predetermined optical structure, mechanical focusing structures mainly adjust the focal length by changing the lens position or the distance between lens groups, and back-end image processing algorithms mainly improve the acquired image data through sharpening, deblurring, brightness correction, or image fusion. However, these methods still suffer from insufficient adjustment precision, weak regional adaptability, and unstable image quality improvement when faced with scenarios involving local defocus, micro-regional aberrations, edge distortion, insufficient local beam convergence, and significant quality differences between different imaging areas.

[0003] A liquid crystal microlens array (LCD) is an arrayed optical device that utilizes electrically controlled changes in the refractive index of liquid crystal materials to achieve optical adjustment. Compared to ordinary fixed microlens arrays, LCDs can change the orientation of liquid crystal molecules within the microlens units through a driving voltage, thereby affecting the refractive and phase states of the corresponding microlens units and thus enabling localized optical path control of the incident imaging beam. Since an LCD is formed by multiple microlens units arranged in an array configuration, it can theoretically implement zoned adjustment for different imaging regions, making it more suitable for improving image quality in scenarios with localized imaging quality deviations. However, current applications of LCDs are mostly concentrated in zooming, beam shaping, light field acquisition, or display modulation. They typically control the LCD array using a uniform driving voltage or a simple regional driving method, making it difficult to perform targeted control of each microlens unit based on the quality deviations of different micro-regions in the initial image data. Summary of the Invention

[0004] One objective of this invention is to propose an image quality improvement method and system based on a liquid crystal microlens array. This invention fully utilizes the electro-optical adjustment capabilities of the liquid crystal microlens array, the microlens response rasterization processing, and the improved quality deviation identification capability of the iDETEX model. It possesses advantages such as high accuracy in image quality identification, highly targeted control of the liquid crystal microlens array, good improvement effects on local defocus and distortion, and stable generation of high-quality target images.

[0005] An image quality improvement method based on a liquid crystal microlens array according to an embodiment of the present invention includes the following steps: The incident imaging beam of the target to be imaged is acquired, the incident imaging beam is guided into the liquid crystal microlens array, and the initial image data formed by the liquid crystal microlens array is acquired in the initial driving state. The initial image data is subjected to microlens response rasterization processing to generate a set of microlens imaging rasteres, and the regional imaging quality characterization parameters of each microlens imaging raster are extracted to generate a set of image quality parameters. The set of image quality parameters is input into the improved iDETEX model to identify quality deviations and generate a set of quality deviation data. Based on the mass deviation data set and the liquid crystal microlens array, the microlens imaging grid corresponding to each mass deviation is array-corresponded with each liquid crystal microlens unit to generate array correspondence data. Based on the array correspondence data, the focal length correction, phase correction and convergence correction for each liquid crystal microlens unit are determined, and a set of liquid crystal microlens array control parameters is generated. Based on the set of control parameters for the liquid crystal microlens array, the driving voltage parameters corresponding to each liquid crystal microlens unit are generated, and the driving matrix of the liquid crystal microlens array is constructed. Based on the driving matrix of the liquid crystal microlens array, a corresponding driving voltage is applied to each liquid crystal microlens unit in the liquid crystal microlens array, and the optical path of the incident imaging beam is controlled to generate controlled image data. The adjusted image data is subjected to focal plane detail fusion processing to generate a high-quality image of the target.

[0006] Optionally, an incident imaging beam formed by reflection or transmission from the target to be imaged is received by an imaging lens, and the incident imaging beam is guided along the imaging optical path to the incident light side of the liquid crystal microlens array. The liquid crystal microlens array is composed of liquid crystal microlens units arranged according to the array position. Before acquiring the initial image data, an initial driving voltage is applied to the liquid crystal microlens array, and each liquid crystal microlens unit is in the initial driving state. In the initial driving state, the imaging beam modulated by the liquid crystal microlens array is received by an image sensor, and the initial image data corresponding to the target to be imaged is acquired. The initial image data characterizes the imaging state of the liquid crystal microlens array before quality compensation adjustment.

[0007] Optionally, the generation of the image quality parameter set specifically includes: The initial image data is subjected to microlens response rasterization processing to generate a set of microlens imaging rasteres. Based on the microlens imaging grid set, regional imaging quality characterization parameters of the microlens imaging grid are extracted, and a set of regional imaging quality characterization parameters is generated. The set of regional imaging quality characterization parameters is arranged in an array to generate a set of image quality parameters.

[0008] Optionally, the generation of the quality deviation data set specifically includes: The improved iDETEX model is used to identify quality deviations by inputting a set of image quality parameters. The improved iDETEX model includes a microlens quality response folding module, a liquid crystal phase degradation sensing module, a cross-grid deviation correlation localization module, and a quality deviation attribution output module. The improvement of the improved iDETEX model is as follows: The traditional iDETEX model uses local image regions as input objects and uses a multimodal image quality evaluation method to locate, perceive, and describe image quality and output image quality evaluation results. The improved iDETEX model uses a set of image quality parameters as input objects, uses a microlens imaging grid as a processing unit, performs array neighborhood response folding processing on the imaging quality characterization parameters of each region, determines the liquid crystal phase degradation characteristics through a response ridge locking mechanism, and generates a quality deviation data set using cross-grid chain tracking and deviation attribution judgment. In the microlens quality response folding module, array neighborhood response folding processing is performed on the imaging quality characterization parameters of each region in the image quality parameter set. The regional imaging quality characterization parameters of the same microlens imaging grid are directionally correlated and compressed with the regional imaging quality characterization parameters of adjacent microlens imaging grids to generate microlens quality response features. In the liquid crystal phase degradation sensing module, a response ridge locking mechanism is introduced. Based on the microlens quality response characteristics, the quality response changes between adjacent microlens imaging grids are tracked along the row direction, column direction and diagonal direction of the microlens imaging grid set, respectively. Ridge locking is performed on microlens imaging grids with continuous quality response changes and concentrated change amplitudes to generate microlens response ridges. Based on the microlens response ridges, the liquid crystal phase degradation characteristics corresponding to each microlens imaging grid are determined. In the cross-grid deviation correlation positioning module, based on the liquid crystal phase degradation characteristics, the degradation feature continuity relationship between adjacent microlens imaging grids is tracked in a cross-grid chain, and microlens imaging grids with consistent degradation feature continuity direction and continuous degradation feature change amplitude are connected as cross-grid deviation chains, and cross-grid deviation positioning data is generated based on the cross-grid deviation chains. In the quality deviation attribution output module, deviation attribution is determined on the cross-grid deviation positioning data to identify the quality deviation type, location, and degree of each microlens imaging grid, and a quality deviation data set is generated.

[0009] Optionally, the generation of the array correspondence data specifically includes: Read the location of the quality deviation in the quality deviation data set, determine the microlens imaging grid with quality deviation, and generate a quality deviation grid set. Read the array position of each liquid crystal microlens unit in the liquid crystal microlens array and generate a set of liquid crystal microlens unit positions; Based on the set of mass deviation grids and the set of liquid crystal microlens unit positions, each mass deviation grid is arrayed with its corresponding liquid crystal microlens unit to generate array correspondence data.

[0010] Optionally, the generation of the set of control parameters for the liquid crystal microlens array specifically includes: Based on the array correspondence data, determine the deviation correction reference data corresponding to each quality deviation; Based on the deviation correction reference data and the array position of the corresponding liquid crystal microlens unit, the focal length correction, phase correction and convergence correction of each mass deviation corresponding to the liquid crystal microlens unit are determined, and a set of liquid crystal microlens array control parameters is generated.

[0011] Optionally, the construction of the liquid crystal microlens array driving matrix specifically includes: Based on the set of control parameters for the liquid crystal microlens array, the focal length driving voltage component of the corresponding liquid crystal microlens unit is determined by the focal length correction amount, the phase driving voltage component of the corresponding liquid crystal microlens unit is determined by the phase correction amount, and the convergence driving voltage component of the corresponding liquid crystal microlens unit is determined by the convergence correction amount. Based on the focal length driving voltage component, phase driving voltage component and convergence driving voltage component, the driving voltage parameters corresponding to each liquid crystal microlens unit are determined. According to the array position of each liquid crystal microlens unit, the driving voltage parameters corresponding to each liquid crystal microlens unit are written into the corresponding array position to construct the liquid crystal microlens array driving matrix.

[0012] Optionally, the generation of the modulated image data specifically includes: According to the array position of each liquid crystal microlens unit in the liquid crystal microlens array, the driving voltage parameters in the liquid crystal microlens array driving matrix are applied to the corresponding liquid crystal microlens unit. Based on the driving voltage parameters, the orientation of the liquid crystal molecules in the corresponding liquid crystal microlens unit is adjusted to form the refractive index distribution and phase distribution; Based on the refractive index distribution and phase distribution, the optical path of the incident imaging beam entering the liquid crystal microlens array is controlled to generate a controlled imaging beam and form controlled image data.

[0013] Optionally, the image data after modulation is read according to the arrangement order of each microlens imaging grid in the microlens imaging grid set to obtain the modulated grid image data corresponding to each microlens imaging grid. The focal plane detail fusion processing is based on the edge change intensity, gray level change intensity, and blur degree in each modulated grid image data to determine the effective detail region in each modulated grid image data. The effective detail region is written into the target image coordinate position according to the arrangement order of the microlens imaging grid set. When adjacent modulated grid image data overlap at the target image coordinate position, the modulated grid image data with high edge change intensity and low blur degree is selected as the image data of the target image coordinate position to generate a high-quality target image.

[0014] An image quality improvement system based on a liquid crystal microlens array according to an embodiment of the present invention includes: The initial image acquisition module is used to acquire the incident imaging beam of the target to be imaged, guide the incident imaging beam into the liquid crystal microlens array, and acquire the initial image data formed by the liquid crystal microlens array in the initial driving state. The image quality parameter generation module is used to perform microlens response rasterization processing on the initial image data, generate a set of microlens imaging rasteres, and extract the regional imaging quality characterization parameters of each microlens imaging raster to generate a set of image quality parameters. The quality deviation identification module is used to input the set of image quality parameters into the improved iDETEX model to identify quality deviations and generate a quality deviation data set. The array correspondence module is used to perform array correspondence between the microlens imaging grid corresponding to each mass deviation and each liquid crystal microlens unit based on the mass deviation data set and the liquid crystal microlens array, and generate array correspondence relationship data. The control parameter generation module is used to determine the focal length correction, phase correction and convergence correction for each liquid crystal microlens unit based on the array correspondence data, and generate a set of control parameters for the liquid crystal microlens array. The driving matrix construction module is used to generate driving voltage parameters corresponding to each liquid crystal microlens unit based on the set of control parameters of the liquid crystal microlens array, and to construct the driving matrix of the liquid crystal microlens array. The optical path control module is used to apply corresponding driving voltages to each liquid crystal microlens unit in the liquid crystal microlens array based on the liquid crystal microlens array driving matrix, and to control the optical path of the incident imaging beam to generate controlled image data. The image fusion output module is used to perform focal plane detail fusion processing on the adjusted image data to generate a high-quality target image.

[0015] The beneficial effects of this invention are: This invention combines a liquid crystal microlens array with an image quality recognition process, enabling image quality improvement to no longer rely solely on back-end image enhancement processing. Instead, it first generates a set of microlens imaging grids and an image quality parameter set based on the initial image data. Then, it uses an improved iDETEX model to identify quality deviations in each microlens imaging grid, providing data for subsequent control of the liquid crystal microlens unit. Therefore, this invention can perform targeted control of the incident imaging beam from the imaging front end, reducing the problems of noise amplification, edge distortion, and insufficient local detail recovery caused by simple back-end enhancement.

[0016] This invention utilizes the array correspondence between a quality deviation data set and a liquid crystal microlens array to accurately map the microlens imaging grids exhibiting quality deviations to the liquid crystal microlens units, thus making the control targets of the liquid crystal microlens array more clearly defined. Based on this array correspondence data, the focal length correction, phase correction, and convergence correction amounts corresponding to each liquid crystal microlens unit can be further determined, and a set of control parameters for the liquid crystal microlens array can be generated. This allows different liquid crystal microlens units to be differentiated according to the quality deviation of their corresponding regions, thereby improving the targeted correction of problems such as local defocusing, phase deviation, brightness unevenness, edge degradation, and distortion shift.

[0017] This invention applies a corresponding driving voltage to each liquid crystal microlens unit via a driving matrix of a liquid crystal microlens array. This causes each liquid crystal microlens unit to form a corresponding refractive index distribution and phase distribution according to the driving voltage parameters, and to control the optical path of the incident imaging beam. This improves the fundamental imaging quality of the controlled image data during the image acquisition stage. Subsequently, focal plane detail fusion processing is performed on the controlled image data, further integrating effective detail regions based on the front-end optical control to generate a high-quality target image, thereby improving image sharpness, edge detail retention, local imaging consistency, and overall image quality stability. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is an overall flowchart of an image quality improvement method and system based on a liquid crystal microlens array proposed in this invention; Figure 2 This is a schematic diagram of the improved iDETEX model of the image quality improvement method and system based on liquid crystal microlens array proposed in this invention; Figure 3 This is a schematic diagram illustrating the construction of the driving matrix of the liquid crystal microlens array, which is a method and system for improving image quality based on a liquid crystal microlens array proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-3 A method for improving image quality based on a liquid crystal microlens array includes the following steps: The incident imaging beam of the target to be imaged is acquired, the incident imaging beam is guided into the liquid crystal microlens array, and the initial image data formed by the liquid crystal microlens array is acquired in the initial driving state. The initial image data is subjected to microlens response rasterization processing to generate a set of microlens imaging rasteres, and the regional imaging quality characterization parameters of each microlens imaging raster are extracted to generate a set of image quality parameters. The set of image quality parameters is input into the improved iDETEX model to identify quality deviations and generate a set of quality deviation data. Based on the mass deviation data set and the liquid crystal microlens array, the microlens imaging grid corresponding to each mass deviation is array-corresponded with each liquid crystal microlens unit to generate array correspondence data. Based on the array correspondence data, the focal length correction, phase correction and convergence correction for each liquid crystal microlens unit are determined, and a set of liquid crystal microlens array control parameters is generated. Based on the set of control parameters for the liquid crystal microlens array, the driving voltage parameters corresponding to each liquid crystal microlens unit are generated, and the driving matrix of the liquid crystal microlens array is constructed. Based on the driving matrix of the liquid crystal microlens array, a corresponding driving voltage is applied to each liquid crystal microlens unit in the liquid crystal microlens array, and the optical path of the incident imaging beam is controlled to generate controlled image data. The adjusted image data is subjected to focal plane detail fusion processing to generate a high-quality image of the target.

[0021] In this embodiment, an incident imaging beam formed by reflection or transmission from the target to be imaged is received by an imaging lens, and the incident imaging beam is guided along the imaging optical path to the light-incident side of the liquid crystal microlens array. The liquid crystal microlens array is composed of liquid crystal microlens units arranged according to the array position. Before acquiring initial image data, an initial driving voltage is applied to the liquid crystal microlens array, and each liquid crystal microlens unit is in the initial driving state. In the initial driving state, the imaging beam modulated by the liquid crystal microlens array is received by an image sensor, and the initial image data corresponding to the target to be imaged is acquired. The initial image data characterizes the imaging state of the liquid crystal microlens array before quality compensation adjustment.

[0022] In this embodiment, the generation of the image quality parameter set specifically includes: The initial image data is subjected to microlens response rasterization processing to generate a set of microlens imaging rasteres. The specific process of generating the microlens imaging grid set is as follows: read the pixel coordinate distribution of the initial image data and obtain the array position of each liquid crystal microlens unit in the liquid crystal microlens array; determine the corresponding imaging response range in the initial image data according to the array position of each liquid crystal microlens unit; perform rasterization segmentation on the initial image data according to each imaging response range, and group the image pixels in the same imaging response range into the same microlens imaging grid, so that each microlens imaging grid corresponds to a liquid crystal microlens unit; summarize all microlens imaging grids to generate the microlens imaging grid set. Based on the microlens imaging grid set, regional imaging quality characterization parameters of the microlens imaging grid are extracted, and a set of regional imaging quality characterization parameters is generated. The specific process for generating the set of regional imaging quality characterization parameters is as follows: Each microlens imaging grid in the microlens imaging grid set is read sequentially, and the image pixels, pixel grayscale values, pixel brightness values, and pixel coordinates within each microlens imaging grid are obtained; the intensity of grayscale changes is calculated based on the pixel grayscale values, obtained by comparing the differences in pixel grayscale values ​​between adjacent image pixels; the difference in brightness distribution is calculated based on the pixel brightness values, obtained by calculating the difference between the pixel brightness values ​​of each image pixel within the same microlens imaging grid and the average pixel brightness value of the microlens imaging grid; the intensity of edge changes is calculated based on pixel coordinates and pixel grayscale values, obtained by determining the magnitude of pixel grayscale value changes at adjacent pixel coordinates; based on grayscale changes… The degree of blurring is calculated based on the intensity and edge change intensity. The degree of blurring is obtained by statistically analyzing the proportion of image pixels where the intensity of gray-level changes and edge changes decreases. The distortion offset is calculated based on the pixel coordinates and the array position of the corresponding liquid crystal microlens unit. The distortion offset is obtained by comparing the positional offset between the pixel coordinate distribution of image pixels within the same microlens imaging grid and the array position of the corresponding liquid crystal microlens unit. The intensity of gray-level changes, brightness distribution differences, edge change intensity, blurring degree, and distortion offset are used as regional imaging quality characterization parameters for the corresponding microlens imaging grid. The regional imaging quality characterization parameters of each microlens imaging grid are summarized according to the arrangement order of each microlens imaging grid in the microlens imaging grid set to generate a set of regional imaging quality characterization parameters. The set of regional imaging quality characterization parameters is arrayed to generate a set of image quality parameters. The specific process of generating the image quality parameter set is as follows: according to the correspondence between each microlens imaging grid in the microlens imaging grid set and each liquid crystal microlens unit in the liquid crystal microlens array, the imaging quality characterization parameters of each region in the region imaging quality characterization parameter set are read, and each region imaging quality characterization parameter is written into the array position of the corresponding liquid crystal microlens unit; only the region imaging quality characterization parameters corresponding to the microlens imaging grid of the same liquid crystal microlens unit are written, and the region imaging quality characterization parameters corresponding to different liquid crystal microlens units are arranged sequentially according to the array row and column order of the liquid crystal microlens units; after completing the writing of all region imaging quality characterization parameters, an image quality parameter set consistent with the array position of the liquid crystal microlens array is generated.

[0023] In this embodiment, the generation of the quality deviation data set specifically includes: The improved iDETEX model is used to identify quality deviations by inputting a set of image quality parameters. The improved iDETEX model includes a microlens quality response folding module, a liquid crystal phase degradation sensing module, a cross-grid deviation correlation localization module, and a quality deviation attribution output module. The improvement of the improved iDETEX model is as follows: The traditional iDETEX model uses local image regions as input objects and uses a multimodal image quality evaluation method to locate, perceive, and describe image quality and output image quality evaluation results. The improved iDETEX model uses a set of image quality parameters as input objects, uses a microlens imaging grid as a processing unit, performs array neighborhood response folding processing on the imaging quality characterization parameters of each region, determines the liquid crystal phase degradation characteristics through a response ridge locking mechanism, and generates a quality deviation data set using cross-grid chain tracking and deviation attribution judgment. The improved iDETEX model has a sequential connection structure. The image quality parameter set is input in the form of an arrayed parameter tensor. The arrayed parameter tensor has the dimension of number of rows × number of columns × number of parameters. The number of rows and columns correspond to the number of rows and columns of the liquid crystal microlens array, respectively, and the number of parameters corresponds to the number of regional imaging quality characterization parameters for each microlens imaging grid. The image quality parameter set is first input to the microlens quality response folding module to generate microlens quality response features with the dimension maintained as number of rows × number of columns × number of features. The microlens quality response features are then input to the liquid crystal phase degradation sensing module to generate liquid crystal phase degradation features with the dimension maintained as number of rows × number of columns × number of degradation features. The liquid crystal phase degradation features are input to the cross-grid deviation correlation localization module to generate cross-grid deviation localization data including cross-grid deviation chain, link grid position, link degradation direction, and link degradation intensity. The cross-grid deviation localization data is input to the quality deviation attribution output module, which outputs a quality deviation data set including quality deviation type, quality deviation location, and quality deviation degree. The training data for the improved iDETEX model comes from the initial image data collected by the liquid crystal microlens array under different initial driving states, different incident imaging beams, and different imaging targets. The data is preprocessed to generate a set of image quality parameters. The training data is labeled as follows: based on the results of manual review and comparison with standard clear images, the quality deviation type, quality deviation location, and quality deviation degree are labeled for each microlens imaging grid. The loss function of the improved iDETEX model is composed of a weighted sum of deviation type loss, deviation location loss, and deviation degree loss. The deviation type loss adopts cross-entropy loss, the deviation location loss adopts grid position intersection-union ratio loss, and the deviation degree loss adopts mean absolute error loss. The overall loss function is expressed as: the total loss is equal to the weighted sum of deviation type loss, deviation location loss, and deviation degree loss. The training parameters were set as follows: batch size was 16, initial learning rate was 0.0001, weight decay coefficient was 0.00001, number of training epochs was 200, the optimizer used was the adaptive moment estimation optimizer, and the learning rate decayed to 0.5 of the previous learning rate every 50 epochs. The convergence condition was set as follows: training stopped when the total loss of the validation set decreased by less than 0.001 for 10 consecutive epochs, or when the accuracy of the validation set quality deviation type identification no longer improved for 10 consecutive epochs. The improved iDETEX model with the lowest total loss of the validation set was saved as the quality deviation identification model. In the microlens quality response folding module, array neighborhood response folding processing is performed on the imaging quality characterization parameters of each region in the image quality parameter set. The regional imaging quality characterization parameters of the same microlens imaging grid are directionally correlated and compressed with the regional imaging quality characterization parameters of adjacent microlens imaging grids to generate microlens quality response features. The generation process of microlens quality response features is as follows: Based on the microlens imaging grid positions corresponding to the imaging quality characterization parameters of each region in the image quality parameter set, the adjacent microlens imaging grids of each microlens imaging grid are determined; the array neighborhood response folding process uses a single microlens imaging grid as the center, and processes the regional imaging quality characterization parameters of the microlens imaging grid with the regional imaging quality characterization parameters of its adjacent microlens imaging grids in the same group, so that the quality changes at the center position and the quality changes at the surrounding positions form the same set of quality response data; directional correlation compression is performed according to the row direction, column direction, and diagonal direction, respectively, on the same set of... The direction and magnitude of change of imaging quality characterization parameters between adjacent regions in the quality response data are compared, and the imaging quality characterization parameters of regions with consistent change direction and continuous change magnitude are compressed into directional quality response components. The row directional quality response components, column directional quality response components, and diagonal directional quality response components corresponding to the same microlens imaging grid are compressed to generate microlens quality response features corresponding to the microlens imaging grid. According to the arrangement order of each microlens imaging grid in the microlens imaging grid set, the microlens quality response features corresponding to each microlens imaging grid are summarized to generate microlens quality response features. In the liquid crystal phase degradation sensing module, a response ridge locking mechanism is introduced. Based on the microlens quality response characteristics, the quality response changes between adjacent microlens imaging grids are tracked along the row direction, column direction and diagonal direction of the microlens imaging grid set, respectively. Ridge locking is performed on microlens imaging grids with continuous quality response changes and concentrated change amplitudes to generate microlens response ridges. Based on the microlens response ridges, the liquid crystal phase degradation characteristics corresponding to each microlens imaging grid are determined. The process for determining the phase degradation characteristics of liquid crystal is as follows: Based on the microlens mass response characteristics, the direction and magnitude of mass response change between adjacent microlens imaging gratings are calculated according to the row, column, and diagonal directions of the microlens imaging grating set, respectively; the response ridge locking mechanism continuously tracks adjacent microlens imaging gratings with consistent mass response change directions and continuously increasing or decreasing mass response change magnitudes in the row, column, and diagonal directions, and determines the continuously tracked microlens imaging gratings as ridge candidate gratings; the ridge candidate gratings are then processed... The ridge line locking process connects candidate gratings located in the same tracking direction and whose quality response change amplitudes satisfy a continuous change relationship into microlens response ridge lines. It reads the microlens quality response characteristics corresponding to each microlens imaging grating in the microlens response ridge lines, and determines the phase degradation direction and phase degradation intensity corresponding to each microlens imaging grating based on the change direction, change amplitude, and continuous distribution position of the microlens quality response characteristics on the microlens response ridge lines. The phase degradation direction and phase degradation intensity are then written into the corresponding microlens imaging grating to generate the liquid crystal phase degradation characteristics corresponding to each microlens imaging grating. In the cross-grid deviation correlation positioning module, based on the liquid crystal phase degradation characteristics, the degradation feature continuity relationship between adjacent microlens imaging grids is tracked in a cross-grid chain, and microlens imaging grids with consistent degradation feature continuity direction and continuous degradation feature change amplitude are connected as cross-grid deviation chains, and cross-grid deviation positioning data is generated based on the cross-grid deviation chains. The specific process for generating cross-grid deviation positioning data is as follows: The phase degradation direction and intensity of each microlens imaging grid in the liquid crystal phase degradation feature are read; according to the row, column, and diagonal directions of the microlens imaging grid set, the phase degradation direction and intensity between adjacent microlens imaging grids are compared. When adjacent microlens imaging grids have the same phase degradation direction and the phase degradation intensity continuously increases or decreases in the same direction, a degradation feature continuity relationship is determined between adjacent microlens imaging grids. The degradation feature continuity relationship means that the phase degradation direction between adjacent microlens imaging grids remains consistent, and the phase... The degradation intensity remains continuously varied between adjacent positions, making adjacent microlens imaging gratings identified as continuous gratings in the same imaging degradation process. Based on the continuity of degradation features, cross-grating chain tracking is performed on microlens imaging gratings that satisfy the continuous connection relationship, and the continuously tracked microlens imaging gratings are connected as cross-grating deviation chains. The grating position, phase degradation direction, and phase degradation intensity of each microlens imaging grating in the cross-grating deviation chain are read, and the grating position is used as the link grating position, the phase degradation direction as the link degradation direction, and the phase degradation intensity as the link degradation intensity. These are then associated and recorded to generate cross-grating deviation positioning data. In the quality deviation attribution output module, deviation attribution is determined on the cross-grid deviation positioning data to determine the quality deviation type, quality deviation location and quality deviation degree corresponding to each microlens imaging grid, and a quality deviation data set is generated. The specific process for generating the quality deviation dataset is as follows: Based on cross-grid deviation positioning data, the microlens imaging grids where quality deviations occur are determined according to the link grid positions, and the link grid positions are used as the quality deviation positions of the corresponding microlens imaging grids; the quality deviation type is determined according to the link degradation direction. When the link degradation direction points to the direction of decreasing grayscale intensity, the quality deviation type of the corresponding microlens imaging grid is determined as sharpness deviation; when the link degradation direction points to the direction of increasing brightness distribution differences, the quality deviation type of the corresponding microlens imaging grid is determined as brightness deviation; when the link degradation direction points to the direction of decreasing edge intensity, the quality deviation type of the corresponding microlens imaging grid is determined as... The deviation type is determined as edge deviation. When the link degradation direction points to the direction of increasing ambiguity, the quality deviation type of the corresponding microlens imaging grid is determined as ambiguity deviation. When the link degradation direction points to the direction of increasing distortion offset, the quality deviation type of the corresponding microlens imaging grid is determined as distortion deviation. The quality deviation degree of the corresponding microlens imaging grid is determined according to the link degradation intensity. The quality deviation degree is determined by the magnitude of the link degradation intensity in the same cross-grid deviation chain. The quality deviation type, quality deviation location, and quality deviation degree corresponding to the same microlens imaging grid are associated and bound, and then collected according to the arrangement order of the cross-grid deviation chain to generate a quality deviation data set.

[0024] In this embodiment, the generation of array correspondence data specifically includes: Read the location of the quality deviation in the quality deviation data set, determine the microlens imaging grid with quality deviation, and generate a quality deviation grid set. The process of generating the quality deviation grid set is as follows: Based on the quality deviation positions in the quality deviation data set, the quality deviation positions are matched with the grid positions of each microlens imaging grid in the microlens imaging grid set; when the quality deviation position falls within the grid range of the microlens imaging grid, the microlens imaging grid is identified as having a quality deviation; when the same microlens imaging grid corresponds to multiple quality deviation positions, the microlens imaging grid is retained once, and the quality deviation information corresponding to the quality deviation position is included in the microlens imaging grid; after completing the matching of all quality deviation positions with the microlens imaging grids, the microlens imaging grids with quality deviations are collected according to their arrangement order in the microlens imaging grid set to generate the quality deviation grid set; Read the array position of each liquid crystal microlens unit in the liquid crystal microlens array and generate a set of liquid crystal microlens unit positions; Based on the set of mass deviation grids and the set of liquid crystal microlens unit positions, each mass deviation grid is arrayed with the corresponding liquid crystal microlens unit to generate array correspondence data; The specific process for generating array correspondence data is as follows: The grid position of each quality deviation grid in the quality deviation grid set is read, and the array position of each liquid crystal microlens unit in the liquid crystal microlens unit position set is read; according to the row and column correspondence between the microlens imaging grid set and the liquid crystal microlens array, the grid position of each quality deviation grid is matched with the array position of the corresponding liquid crystal microlens unit; when the grid row number and grid column number of the quality deviation grid are consistent with the array row number and array column number of the liquid crystal microlens unit, the quality deviation grid and the liquid crystal microlens unit are determined to have an array correspondence; based on the quality deviation type, quality deviation position, and quality deviation degree corresponding to the quality deviation grid in the quality deviation data set, the array position, quality deviation type, quality deviation position, and quality deviation degree of the quality deviation grid and the liquid crystal microlens unit are associated and recorded to generate array correspondence data.

[0025] In this embodiment, the generation of the set of control parameters for the liquid crystal microlens array specifically includes: Based on the array correspondence data, determine the deviation correction reference data corresponding to each quality deviation; The process of determining the deviation correction reference data is as follows: Read the quality deviation type, degree, and array position of the corresponding liquid crystal microlens unit for each quality deviation in the array correspondence data; determine the correction object corresponding to the quality deviation based on the quality deviation type. When the quality deviation type is sharpness deviation or blurring deviation, the corresponding correction object is determined as the focal length deviation reference value; when the quality deviation type is edge deviation or distortion deviation, the corresponding correction object is determined as the phase deviation reference value; when the quality deviation type is brightness deviation, the corresponding correction object is determined as the convergence deviation reference value; determine the reference value of the corresponding correction object based on the degree of quality deviation, so that the degree of quality deviation and the reference value of the corresponding correction object form a positive correspondence; associate and record the focal length deviation reference value, phase deviation reference value, and convergence deviation reference value with the array position of the corresponding liquid crystal microlens unit to generate the deviation correction reference data corresponding to each quality deviation. Based on the deviation correction reference data and the array position of the corresponding liquid crystal microlens unit, the focal length correction, phase correction and convergence correction of each quality deviation corresponding to the liquid crystal microlens unit are determined, and a set of liquid crystal microlens array control parameters is generated. The specific process for generating the set of control parameters for the liquid crystal microlens array is as follows: Read the focal length deviation reference value, phase deviation reference value, and convergence deviation reference value corresponding to each quality deviation in the deviation correction reference data, as well as the array position of the corresponding liquid crystal microlens unit; multiply the focal length deviation reference value by the degree of quality deviation of the corresponding quality deviation to obtain the focal length correction value for the corresponding liquid crystal microlens unit, which represents the magnitude of focal length change that needs to be corrected for the corresponding liquid crystal microlens unit; multiply the phase deviation reference value by the degree of quality deviation of the corresponding quality deviation to obtain the phase correction value for the corresponding liquid crystal microlens unit, which represents the magnitude of phase change that needs to be corrected for the corresponding liquid crystal microlens unit; multiply the convergence deviation reference value by the degree of quality deviation of the corresponding quality deviation to obtain the convergence correction value for the corresponding liquid crystal microlens unit, which represents the magnitude of beam convergence change that needs to be corrected for the corresponding liquid crystal microlens unit; associate and record the array position of each liquid crystal microlens unit with the corresponding focal length correction value, phase correction value, and convergence correction value to generate the set of control parameters for the liquid crystal microlens array.

[0026] In this embodiment, the construction of the liquid crystal microlens array driving matrix specifically includes: Based on the set of control parameters for the liquid crystal microlens array, the focal length driving voltage component of the corresponding liquid crystal microlens unit is determined by the focal length correction amount, the phase driving voltage component of the corresponding liquid crystal microlens unit is determined by the phase correction amount, and the convergence driving voltage component of the corresponding liquid crystal microlens unit is determined by the convergence correction amount. The determination process for the focal length driving voltage component, phase driving voltage component, and convergence driving voltage component is as follows: Read the focal length correction, phase correction, and convergence correction values ​​corresponding to each liquid crystal microlens unit in the set of control parameters for the liquid crystal microlens array, and read the array position of each liquid crystal microlens unit; based on the voltage response relationship of the liquid crystal microlens unit, convert the focal length correction value into a focal length driving voltage component, the phase correction value into a phase driving voltage component, and the convergence correction value into a convergence driving voltage component; the voltage response relationship is formed by the change in liquid crystal molecule orientation, refractive index distribution, and phase distribution of the liquid crystal microlens unit under different driving voltages. The correspondence is as follows: the focal length driving voltage component is used to change the equivalent focal length of the liquid crystal microlens unit; the phase driving voltage component is used to change the phase distribution of the liquid crystal microlens unit; and the convergence driving voltage component is used to change the convergence state of the liquid crystal microlens unit for the incident imaging beam. During the conversion, the focal length correction amount is compared with the focal length change caused by the unit voltage of the liquid crystal microlens unit to obtain the focal length driving voltage component; the phase correction amount is compared with the phase change caused by the unit voltage of the liquid crystal microlens unit to obtain the phase driving voltage component; and the convergence correction amount is compared with the convergence change caused by the unit voltage of the liquid crystal microlens unit to obtain the convergence driving voltage component. Based on the focal length driving voltage component, phase driving voltage component and convergence driving voltage component, the driving voltage parameters corresponding to each liquid crystal microlens unit are determined. The process of determining the driving voltage parameters is as follows: Based on the focal length driving voltage component, phase driving voltage component, and convergence driving voltage component corresponding to each liquid crystal microlens unit, voltage components are synthesized according to the array position of each liquid crystal microlens unit. Voltage component synthesis involves merging the focal length driving voltage component, phase driving voltage component, and convergence driving voltage component corresponding to the same liquid crystal microlens unit according to the voltage direction. When the three voltage components have the same direction, they are added together to obtain the driving voltage amplitude of the liquid crystal microlens unit. When any voltage component has the opposite direction, it is subtracted from the driving voltage amplitude as a reverse correction. The driving voltage duration is determined according to the degree of quality deviation corresponding to the liquid crystal microlens unit. The greater the quality deviation, the longer the driving voltage duration. The driving voltage amplitude and driving voltage duration are used as the driving voltage parameters corresponding to the liquid crystal microlens unit. According to the array position of each liquid crystal microlens unit, the driving voltage parameters corresponding to each liquid crystal microlens unit are written into the corresponding array position to construct the liquid crystal microlens array driving matrix. The construction process of the liquid crystal microlens array driving matrix is ​​as follows: based on the array position of each liquid crystal microlens unit and the corresponding driving voltage parameters; according to the row and column positions of each liquid crystal microlens unit in the liquid crystal microlens array, establish a voltage writing table consistent with the number of rows and columns of the liquid crystal microlens array; write the driving voltage amplitude and driving voltage duration corresponding to each liquid crystal microlens unit into the cell of the same array position in the voltage writing table; when the liquid crystal microlens unit does not correspond to the quality deviation, write the driving voltage parameters in the initial driving state into the cell corresponding to the liquid crystal microlens unit; after completing the writing of the driving voltage parameters of all liquid crystal microlens units, use the voltage writing table as the liquid crystal microlens array driving matrix.

[0027] In this embodiment, the generation of the adjusted image data specifically includes: According to the array position of each liquid crystal microlens unit in the liquid crystal microlens array, the driving voltage parameters in the liquid crystal microlens array driving matrix are applied to the corresponding liquid crystal microlens unit. Based on the driving voltage parameters, the orientation of the liquid crystal molecules in the corresponding liquid crystal microlens unit is adjusted to form the refractive index distribution and phase distribution; The formation process of the refractive index distribution and phase distribution is as follows: a driving voltage parameter is applied to the electrode layer of the corresponding liquid crystal microlens unit, forming an electric field distribution between the electrode layers; the liquid crystal molecules in the liquid crystal microlens unit undergo orientation deflection according to the electric field distribution, forming a liquid crystal molecule orientation distribution; based on the liquid crystal molecule orientation distribution, the equivalent refractive index at different positions in the liquid crystal microlens unit changes, forming a refractive index distribution; when the incident imaging beam passes through the refractive index distribution, different beam positions produce different optical path changes, and the optical path changes are continuously distributed according to the spatial position of the liquid crystal microlens unit, forming a phase distribution; Based on the refractive index distribution and phase distribution, the incident imaging beam entering the liquid crystal microlens array is optically controlled to generate a controlled imaging beam and form controlled image data. The specific process of forming the image data after modulation is as follows: After the incident imaging beam enters the liquid crystal microlens array, it passes through the refractive index distribution and phase distribution formed by each liquid crystal microlens unit in sequence; the optical path modulation changes the propagation direction of the incident imaging beam in each liquid crystal microlens unit by changing the refractive index distribution, and adjusts the phase delay of the incident imaging beam at different beam positions by adjusting the phase distribution, so that the incident imaging beam passing through the same liquid crystal microlens unit forms the modulated imaging beam according to the corresponding focal length correction, phase correction and convergence correction; after the modulated imaging beam is incident on the photosensitive area of ​​the image sensor, a light intensity distribution corresponding to each microlens imaging grid is formed in the photosensitive area; the image sensor generates pixel response values ​​according to the light intensity distribution received by each pixel position in the photosensitive area, and records the pixel response values ​​of each pixel position according to the pixel coordinates to form the modulated image data.

[0028] In this embodiment, the image data after modulation is read according to the arrangement order of each microlens imaging grid in the microlens imaging grid set to obtain the modulated grid image data corresponding to each microlens imaging grid. The focal plane detail fusion processing is based on the edge change intensity, gray scale change intensity, and blur degree in each modulated grid image data to determine the effective detail region in each modulated grid image data. The effective detail region is written into the target image coordinate position according to the arrangement order of the microlens imaging grid set. When adjacent modulated grid image data overlap at the target image coordinate position, the modulated grid image data with high edge change intensity and low blur degree is selected as the image data of the target image coordinate position to generate a high-quality target image.

[0029] An image quality improvement system based on a liquid crystal microlens array includes: The initial image acquisition module is used to acquire the incident imaging beam of the target to be imaged, guide the incident imaging beam into the liquid crystal microlens array, and acquire the initial image data formed by the liquid crystal microlens array in the initial driving state. The image quality parameter generation module is used to perform microlens response rasterization processing on the initial image data, generate a set of microlens imaging rasteres, and extract the regional imaging quality characterization parameters of each microlens imaging raster to generate a set of image quality parameters. The quality deviation identification module is used to input the set of image quality parameters into the improved iDETEX model to identify quality deviations and generate a quality deviation data set. The array correspondence module is used to perform array correspondence between the microlens imaging grid corresponding to each mass deviation and each liquid crystal microlens unit based on the mass deviation data set and the liquid crystal microlens array, and generate array correspondence relationship data. The control parameter generation module is used to determine the focal length correction, phase correction and convergence correction for each liquid crystal microlens unit based on the array correspondence data, and generate a set of control parameters for the liquid crystal microlens array. The driving matrix construction module is used to generate driving voltage parameters corresponding to each liquid crystal microlens unit based on the set of control parameters of the liquid crystal microlens array, and to construct the driving matrix of the liquid crystal microlens array. The optical path control module is used to apply corresponding driving voltages to each liquid crystal microlens unit in the liquid crystal microlens array based on the liquid crystal microlens array driving matrix, and to control the optical path of the incident imaging beam to generate controlled image data. The image fusion output module is used to perform focal plane detail fusion processing on the adjusted image data to generate a high-quality target image.

[0030] Example 1: This invention is applied to a miniature industrial vision inspection device used to detect minute scratches, edge burrs, and localized depressions on the surface of transparent plastic microstructures. Because the surface of the workpiece has slight curvature variations, and the workpiece's height fluctuates slightly as it passes through the imaging area on the conveyor belt, images acquired by ordinary fixed-focus lenses are prone to localized defocusing, blurred edges, uneven brightness distribution, and localized distortion. If only post-processing sharpening or deblurring algorithms are used, problems such as excessive edge enhancement, misidentification of fine scratches as noise, and discontinuous local textures can easily occur, affecting the accuracy of subsequent defect identification.

[0031] In this scenario, a liquid crystal microlens array is positioned between the imaging lens and the image sensor. When the detection device is operating, the imaging lens receives the incident imaging beam reflected from the target object and guides it into the liquid crystal microlens array. In the initial driving state, the image sensor acquires initial image data formed by the liquid crystal microlens array. In this initial image data, there is a significant decrease in edge variation intensity near the workpiece edge, increased brightness distribution differences in local reflective areas, and decreased grayscale variation intensity and increased blurring in some fine texture areas.

[0032] The initial image data undergoes microlens response rasterization. Based on the array position of each liquid crystal microlens unit in the liquid crystal microlens array, the initial image data is divided into multiple microlens imaging grids, forming a microlens imaging grid set. For each microlens imaging grid, the image pixels, pixel grayscale values, pixel brightness values, and pixel coordinates are read. Grayscale change intensity, brightness distribution difference, edge change intensity, blur degree, and distortion offset are calculated, and these parameters are used as regional imaging quality characterization parameters. The regional imaging quality characterization parameters of each microlens imaging grid are arranged according to the array position of the liquid crystal microlens units to generate an image quality parameter set.

[0033] The image quality parameter set is input into the improved iDETEX model for quality deviation identification. The improved iDETEX model uses a microlens imaging grid as the processing unit. A microlens quality response folding module performs directional correlation compression on quality changes between adjacent microlens imaging grids to generate microlens quality response features. Then, through the response ridge locking mechanism in the liquid crystal phase degradation sensing module, quality response changes are tracked along the row, column, and diagonal directions of the microlens imaging grid set to generate liquid crystal phase degradation features. Finally, a cross-grid deviation correlation localization module generates cross-grid deviation localization data, and a quality deviation attribution output module determines the quality deviation type, location, and degree, generating a quality deviation data set.

[0034] Based on the mass deviation data set and the liquid crystal microlens array, the microlens imaging grid corresponding to each mass deviation is mapped to the liquid crystal microlens unit, generating array mapping data. Based on this mapping data, the focal length correction, phase correction, and convergence correction for each liquid crystal microlens unit are determined, generating a set of liquid crystal microlens array control parameters. Subsequently, driving voltage parameters are generated based on this set of parameters, and a liquid crystal microlens array driving matrix is ​​constructed. The driving matrix is ​​then programmed with the driving voltage parameters according to the array position of each liquid crystal microlens unit, ensuring that different liquid crystal microlens units receive driving voltages that match their corresponding mass deviations.

[0035] During the modulation process, the driving voltage parameters in the driving matrix of the liquid crystal microlens array are applied to the corresponding liquid crystal microlens units, adjusting the orientation of liquid crystal molecules within the units and forming corresponding refractive index and phase distributions. When the incident imaging beam passes through the liquid crystal microlens array, the optical paths corresponding to different microlens imaging grids are modulated separately, thereby forming modulated image data. Finally, focal plane detail fusion processing is performed on the modulated image data. The image data of each modulated grid is read according to the arrangement order of the microlens imaging grid set, and effective detail regions are selected based on edge change intensity, grayscale change intensity, and blur degree to generate a high-quality target image.

[0036] To verify the technical effectiveness of this embodiment, 300 sets of surface inspection images of transparent plastic microstructures were selected for testing. Each set of images included edge regions, fine texture regions, and local reflective regions. The comparison methods included unprocessed initial images, images with only post-processing image enhancement, and high-quality target images processed using the method of this invention. Test indicators included sharpness evaluation value, edge retention rate, brightness uniformity error, proportion of locally blurred regions, and defect identification accuracy. Higher sharpness evaluation values ​​and edge retention rates indicate better image quality, while lower brightness uniformity error and proportion of locally blurred regions indicate better image quality.

[0037] Table 1 Comparison of Imaging Effects of Different Image Quality Improvement Methods

[0038] As shown in Table 1, the unprocessed initial image suffers from local defocusing, insufficient beam convergence, and edge distortion, resulting in a sharpness rating of only 0.62 and a local blurred area ratio of 18.6%, leading to low accuracy in identifying minor scratches and edge burrs. After using only back-end image enhancement processing, the sharpness rating improved to 0.74, and the proportion of local blurred areas decreased to 12.9%. However, since this method does not change the front-end imaging optical path, it is still difficult to fully correct the local phase deviation and convergence deviation generated by the corresponding area of ​​the liquid crystal microlens unit.

[0039] After adopting the method of this invention, the sharpness evaluation value of the target high-quality image is improved to 0.87, the edge preservation rate reaches 91.5%, the brightness uniformity error is reduced to 6.9%, the proportion of locally blurred areas is reduced to 5.8%, and the defect identification accuracy is improved to 96.2%. These results demonstrate that this invention, through microlens response rasterization processing and the improved iDETEX model to identify quality deviation data sets, can accurately map microlens imaging gratings with quality deviations to liquid crystal microlens units. Furthermore, by using a liquid crystal microlens array driving matrix to perform targeted optical path control on the incident imaging beam, the controlled image data already possesses better basic imaging quality before entering the focal plane detail fusion processing. Therefore, this invention can effectively improve problems such as local defocusing, edge blurring, brightness unevenness, and distortion shift, and improve the detail integrity of the target high-quality image and the stability of subsequent defect identification.

[0040] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for improving image quality based on a liquid crystal microlens array, characterized in that, Includes the following steps: The incident imaging beam of the target to be imaged is acquired, the incident imaging beam is guided into the liquid crystal microlens array, and the initial image data formed by the liquid crystal microlens array is acquired in the initial driving state. The initial image data is subjected to microlens response rasterization processing to generate a set of microlens imaging rasteres, and the regional imaging quality characterization parameters of each microlens imaging raster are extracted to generate a set of image quality parameters. The image quality parameter set is input into the improved iDETEX model to identify quality deviations and generate a quality deviation data set. Based on the mass deviation data set and the liquid crystal microlens array, the microlens imaging grid corresponding to each mass deviation is array-corresponded with each liquid crystal microlens unit to generate array correspondence data. Based on the array correspondence data, the focal length correction, phase correction and convergence correction for each liquid crystal microlens unit are determined, and a set of liquid crystal microlens array control parameters is generated. Based on the set of control parameters for the liquid crystal microlens array, the driving voltage parameters corresponding to each liquid crystal microlens unit are generated, and the driving matrix of the liquid crystal microlens array is constructed. Based on the driving matrix of the liquid crystal microlens array, a corresponding driving voltage is applied to each liquid crystal microlens unit in the liquid crystal microlens array, and the optical path of the incident imaging beam is controlled to generate controlled image data. The adjusted image data is subjected to focal plane detail fusion processing to generate a high-quality image of the target.

2. The image quality improvement method based on a liquid crystal microlens array according to claim 1, characterized in that, An imaging lens receives an incident imaging beam formed by reflection or transmission from the target to be imaged, and guides the incident imaging beam along the imaging optical path to the incident light side of a liquid crystal microlens array. The liquid crystal microlens array is composed of liquid crystal microlens units arranged according to the array position. Before acquiring initial image data, an initial driving voltage is applied to the liquid crystal microlens array, and each liquid crystal microlens unit is in the initial driving state. In the initial driving state, the image sensor receives the imaging beam modulated by the liquid crystal microlens array and acquires the initial image data corresponding to the target to be imaged. The initial image data characterizes the imaging state of the liquid crystal microlens array before quality compensation adjustment.

3. The image quality improvement method based on a liquid crystal microlens array according to claim 1, characterized in that, The generation of the image quality parameter set specifically includes: The initial image data is subjected to microlens response rasterization processing to generate a set of microlens imaging rasteres. Based on the microlens imaging grid set, regional imaging quality characterization parameters of the microlens imaging grid are extracted, and a set of regional imaging quality characterization parameters is generated. The set of regional imaging quality characterization parameters is arranged in an array to generate a set of image quality parameters.

4. The image quality improvement method based on a liquid crystal microlens array according to claim 1, characterized in that, The generation of the quality deviation data set specifically includes: The improved iDETEX model is used to identify quality deviations by inputting a set of image quality parameters. The improved iDETEX model includes a microlens quality response folding module, a liquid crystal phase degradation sensing module, a cross-grid deviation correlation localization module, and a quality deviation attribution output module. The improvement of the improved iDETEX model is as follows: The traditional iDETEX model uses local image regions as input objects and uses a multimodal image quality evaluation method to locate, perceive, and describe image quality and output image quality evaluation results. The improved iDETEX model uses a set of image quality parameters as input objects, uses a microlens imaging grid as a processing unit, performs array neighborhood response folding processing on the imaging quality characterization parameters of each region, determines the liquid crystal phase degradation characteristics through a response ridge locking mechanism, and generates a quality deviation data set using cross-grid chain tracking and deviation attribution judgment. In the microlens quality response folding module, array neighborhood response folding processing is performed on the imaging quality characterization parameters of each region in the image quality parameter set. The regional imaging quality characterization parameters of the same microlens imaging grid are directionally correlated and compressed with the regional imaging quality characterization parameters of adjacent microlens imaging grids to generate microlens quality response features. In the liquid crystal phase degradation sensing module, a response ridge locking mechanism is introduced. Based on the microlens quality response characteristics, the quality response changes between adjacent microlens imaging grids are tracked along the row direction, column direction and diagonal direction of the microlens imaging grid set, respectively. Ridge locking is performed on microlens imaging grids with continuous quality response changes and concentrated change amplitudes to generate microlens response ridges. Based on the microlens response ridges, the liquid crystal phase degradation characteristics corresponding to each microlens imaging grid are determined. In the cross-grid deviation correlation positioning module, based on the liquid crystal phase degradation characteristics, the degradation feature continuity relationship between adjacent microlens imaging grids is tracked in a cross-grid chain, and microlens imaging grids with consistent degradation feature continuity direction and continuous degradation feature change amplitude are connected as cross-grid deviation chains, and cross-grid deviation positioning data is generated based on the cross-grid deviation chains. In the quality deviation attribution output module, deviation attribution is determined on the cross-grid deviation positioning data to identify the quality deviation type, location, and degree of each microlens imaging grid, and a quality deviation data set is generated.

5. The image quality improvement method based on a liquid crystal microlens array according to claim 1, characterized in that, The generation of the array correspondence data specifically includes: Read the location of the quality deviation in the quality deviation data set, determine the microlens imaging grid with quality deviation, and generate a quality deviation grid set. Read the array position of each liquid crystal microlens unit in the liquid crystal microlens array and generate a set of liquid crystal microlens unit positions; Based on the set of mass deviation grids and the set of liquid crystal microlens unit positions, each mass deviation grid is arrayed with its corresponding liquid crystal microlens unit to generate array correspondence data.

6. The image quality improvement method based on a liquid crystal microlens array according to claim 1, characterized in that, The generation of the set of control parameters for the liquid crystal microlens array specifically includes: Based on the array correspondence data, determine the deviation correction reference data corresponding to each quality deviation; Based on the deviation correction reference data and the array position of the corresponding liquid crystal microlens unit, the focal length correction, phase correction and convergence correction of each mass deviation corresponding to the liquid crystal microlens unit are determined, and a set of liquid crystal microlens array control parameters is generated.

7. The image quality improvement method based on a liquid crystal microlens array according to claim 1, characterized in that, The construction of the liquid crystal microlens array driving matrix specifically includes: Based on the set of control parameters for the liquid crystal microlens array, the focal length driving voltage component of the corresponding liquid crystal microlens unit is determined by the focal length correction amount, the phase driving voltage component of the corresponding liquid crystal microlens unit is determined by the phase correction amount, and the convergence driving voltage component of the corresponding liquid crystal microlens unit is determined by the convergence correction amount. Based on the focal length driving voltage component, phase driving voltage component and convergence driving voltage component, the driving voltage parameters corresponding to each liquid crystal microlens unit are determined. According to the array position of each liquid crystal microlens unit, the driving voltage parameters corresponding to each liquid crystal microlens unit are written into the corresponding array position to construct the liquid crystal microlens array driving matrix.

8. The image quality improvement method based on a liquid crystal microlens array according to claim 1, characterized in that, The generation of the modulated image data specifically includes: According to the array position of each liquid crystal microlens unit in the liquid crystal microlens array, the driving voltage parameters in the liquid crystal microlens array driving matrix are applied to the corresponding liquid crystal microlens unit. Based on the driving voltage parameters, the orientation of the liquid crystal molecules in the corresponding liquid crystal microlens unit is adjusted to form the refractive index distribution and phase distribution; Based on the refractive index distribution and phase distribution, the optical path of the incident imaging beam entering the liquid crystal microlens array is controlled to generate a controlled imaging beam and form controlled image data.

9. The image quality improvement method based on a liquid crystal microlens array according to claim 1, characterized in that, According to the arrangement order of each microlens imaging grid in the microlens imaging grid set, the modulated image data is read grid-to-grid to obtain the modulated grid image data corresponding to each microlens imaging grid. The focal plane detail fusion processing is based on the edge change intensity, grayscale change intensity, and blur degree in each modulated grid image data to determine the effective detail region in each modulated grid image data. The effective detail region is written into the target image coordinate position according to the arrangement order of the microlens imaging grid set. When adjacent modulated grid image data overlap at the target image coordinate position, the modulated grid image data with high edge change intensity and low blur degree is selected as the image data of the target image coordinate position to generate a high-quality target image.

10. An image quality improvement system based on a liquid crystal microlens array, comprising the image quality improvement method based on a liquid crystal microlens array as described in any one of claims 1 to 9, characterized in that, include: The initial image acquisition module is used to acquire the incident imaging beam of the target to be imaged, guide the incident imaging beam into the liquid crystal microlens array, and acquire the initial image data formed by the liquid crystal microlens array in the initial driving state. The image quality parameter generation module is used to perform microlens response rasterization processing on the initial image data, generate a set of microlens imaging rasteres, and extract the regional imaging quality characterization parameters of each microlens imaging raster to generate a set of image quality parameters. The quality deviation identification module is used to input the set of image quality parameters into the improved iDETEX model to identify quality deviations and generate a quality deviation data set. The array correspondence module is used to perform array correspondence between the microlens imaging grid corresponding to each quality deviation and each liquid crystal microlens unit based on the quality deviation data set and the liquid crystal microlens array, and generate array correspondence relationship data. The control parameter generation module is used to determine the focal length correction, phase correction and convergence correction for each liquid crystal microlens unit based on the array correspondence data, and generate a set of control parameters for the liquid crystal microlens array. The driving matrix construction module is used to generate driving voltage parameters corresponding to each liquid crystal microlens unit based on the set of control parameters of the liquid crystal microlens array, and to construct the driving matrix of the liquid crystal microlens array. The optical path control module is used to apply corresponding driving voltages to each liquid crystal microlens unit in the liquid crystal microlens array based on the liquid crystal microlens array driving matrix, and to control the optical path of the incident imaging beam to generate controlled image data. The image fusion output module is used to perform focal plane detail fusion processing on the adjusted image data to generate a high-quality target image.

Images