Card camera power consumption optimization method and system based on image processing

Abstract

The invention provides a card camera power consumption optimization method and system based on image processing, and relates to the technical field of camera power consumption optimization, and the method comprises the steps: obtaining a lens light field of a card camera; compressing the lens light field to reduce redundant data of the lens light field; according to the compressed lens light field, constructing a photosensitive unit activation matrix of the card camera based on a Kailer geometry algorithm; controlling a light sensing unit of the card camera through the light sensing unit activation matrix, and collecting a live-action image; carrying out single-time conformal mapping on the real-scene image so as to carry out denoising on the real-scene image; with minimization of an entropy output value representing image information loss as a constraint, encoding the de-noised live-action image to obtain a target live-action image; and outputting the target real-scene image. The data is compressed on the premise of ensuring that the image quality is not lost, the high-quality target image is finally output, and the image quality is maintained while the power consumption is remarkably reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of camera power consumption optimization, and in particular to a card camera power consumption optimization method and system based on image processing. Background Art

[0002] A compact camera is a digital camera that is mainly used for daily photography. Compact camera power consumption optimization refers to reducing the power consumed by the camera when taking pictures, processing images, and performing other operations by optimizing the camera hardware and internal image processing logic. Usually, this involves optimizing the use of the camera's sensor and processor, reducing unnecessary computing processes and power consumption, and extending the camera's battery life.

[0003] The necessity of optimizing the power consumption of compact cameras is that with the widespread use of portable devices, users have higher requirements for the battery life of cameras. Traditional cameras have limited battery life due to the need for a lot of computing and high-power sensors. Especially during long-term shooting and high-resolution image processing, the rapid consumption of power seriously affects the user experience. By optimizing power consumption, not only can the battery life be extended, but also the working efficiency of the device can be improved, and the user experience and satisfaction can be enhanced. Therefore, optimizing the power consumption of compact cameras is not only a demand for technological development, but also the key to improving market competitiveness.

[0004] However, existing compact camera power consumption optimization mostly relies on reducing hardware power consumption, such as by adjusting the sampling rate of the sensor or using low power mode, but these methods often affect image quality, especially in the process of high-resolution and high dynamic range image processing, and cannot achieve efficient power consumption control while ensuring image quality. Summary of the invention

[0005] In order to solve the technical problem that the power consumption optimization of card cameras in the prior art mostly relies on reducing the power consumption of hardware, such as by adjusting the sampling rate of the sensor or using a low power consumption mode, but these methods often affect the image quality, especially in the process of high-resolution and high dynamic range image processing, it is impossible to achieve efficient power consumption control while ensuring image quality, the present invention provides a card camera power consumption optimization method and system based on image processing.

[0006] The technical solution provided by the embodiment of the present invention is as follows:

[0007] First aspect An embodiment of the present invention provides a card camera power consumption optimization method based on image processing, comprising: S1: Get the lens light field of the card camera; S2: Compression of the lens light field is performed in combination with the Gromov-Wasserstein distance to reduce redundant data of the lens light field; S3: Based on the compressed lens light field, the photosensitive unit activation matrix of the card camera is constructed based on the Keller geometry algorithm; S4: Control the photosensitive unit of the card camera through the photosensitive unit activation matrix to collect real scene images; S5: performing single conformal mapping on the real scene image to denoise the real scene image; S6: Encode the denoised real scene image under the constraint of minimizing the entropy production value representing the loss of image information to obtain the target real scene image; S7: Output the target real scene image.

[0008] Second aspect An embodiment of the present invention provides a card camera power consumption optimization system based on image processing, comprising: processor; The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the card camera power consumption optimization method based on image processing as in the first aspect is implemented.

[0009] The third aspect An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method for optimizing power consumption of a compact camera based on image processing according to the first aspect is implemented.

[0010] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least: In an embodiment of the present invention, first, the lens light field is obtained to provide complete optical information for subsequent image processing, ensuring the basic data quality of image processing. Then, the lens light field is compressed by the Gromov-Wasserstein distance, which effectively reduces redundant data, reduces power consumption from the source, and avoids the problem of sacrificing image quality to save power in traditional methods. Then, the Keller geometry algorithm is used to construct a photosensitive unit activation matrix, and only the required photosensitive units are activated, further reducing unnecessary calculations and power consumption. Then, the real-scene image captured by the card camera based on the photosensitive unit activation matrix is ​​denoised once through angle-preserving mapping, which can further save power consumption while reducing the impact of noise and improving image quality. Finally, an optimized encoding method based on entropy minimization is used to ensure that data is compressed without losing image quality, and finally a high-quality target image is output, effectively balancing power consumption and image quality. While significantly reducing power consumption, image quality is maintained. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A schematic flow chart of a card camera power consumption optimization method based on image processing provided by an embodiment of the present invention;

[0013] Figure 2 A schematic diagram of the structure of a card camera power consumption optimization system based on image processing provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0015] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0016] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0017] Reference Manual Attached Figure 1 , shows a flow chart of a card camera power consumption optimization method based on image processing provided by an embodiment of the present invention.

[0018] The embodiment of the present invention provides a card camera power consumption optimization method based on image processing, which can be implemented by a card camera power consumption optimization device based on image processing, and the card camera power consumption optimization device based on image processing can be a terminal or a server. The processing flow of the card camera power consumption optimization method based on image processing may include the following steps:

[0019] S1: Get the light field of the camera lens.

[0020] Among them, the lens light field refers to a four-dimensional data structure captured by a light field camera, which describes the propagation and distribution of light in space. Specifically, the light field not only contains information about the spatial position, but also the direction and angle of the light. It can be understood as a high-dimensional data model that describes the behavior of light in three-dimensional space, in which the light at each point is represented by two main parameters: one is the spatial position of the point (x, y, z), and the other is the propagation direction of the light (usually expressed as an angle).

[0021] Alternatively, the compact camera can capture the position and direction of light in space through a specific light field sensor or multi-lens array to obtain four-dimensional light field data. These data record the spatial coordinates and direction angle of each light ray, forming a complete lens light field, providing rich optical information for subsequent image processing.

[0022] S2: The lens light field is compressed in combination with the Gromov-Wasserstein distance to reduce the redundant data of the lens light field.

[0023] Among them, the GW distance (Gromov-Wasserstein distance) is a mathematical tool for measuring the structural differences between two metric spaces. It minimizes the matching cost between them by considering the point sets in the two spaces and the mapping relationship between them. The core of the GW distance is to make the two spaces closest in distance measurement by selecting the most appropriate matching scheme. In image processing, the GW distance can help find the most representative light direction, thereby compressing the light field data and removing redundant light information, thereby reducing the volume of data and reducing power consumption.

[0024] It should be noted that compressing the lens light field through the Gromov-Wasserstein distance can effectively reduce the amount of redundant data in the light field, thereby reducing the computational burden of subsequent processing. This method can select the most representative light direction in the light field manifold, ensure that only the light that contributes to the image quality is retained, avoid the transmission and processing of invalid information, thereby significantly reducing the total amount of data and power consumption while maintaining image quality.

[0025] In a possible implementation, S2 specifically includes:

[0026] S201: Obtain geometric measurements of light rays in the lens light field at different positions and directions, that is, probability distribution.

[0027] It should be noted that by quantifying the spatial distribution characteristics of light in the lens light field, the geometric measurement and probability distribution of light in four-dimensional coordinates (position and direction) can be obtained. Specifically, the light field sensor can be used to record the spatial coordinates (x, y, z) and propagation direction (angle parameter) of the light, and the probability density function of the light field manifold can be constructed by counting the density and direction distribution of light in each area. By accurately quantifying the spatial distribution and direction information of light, redundant data can be effectively reduced, providing high-quality basic data for subsequent light field compression and power consumption optimization, and improving the processing efficiency and energy efficiency of the camera.

[0028] S202: Establishing a matching relationship between the lens light field and the reference lens light field in combination with geometric measurement, and outputting an optimal matching solution between the lens light field and the reference lens light field with the constraint of minimizing the matching cost between the lens light field and the reference lens light field.

[0029] The reference lens light field includes the sparse light structure that is expected to be retained in the design. It is a predefined sparse light field model containing the core optical structure. It forms a low-redundancy geometric template by screening key light directions (such as light at the edge of an object or in a textured area) to guide the compression process to retain optical information that has a significant impact on the image quality.

[0030] The optimal matching solution is calculated as follows: ; ; in, represents the optimal matching solution, Respectively represent the light distribution in the lens light field and the reference lens light field, Indicates compliance The light and the The set of matching solutions between the rays, Represents a matching scheme in a matching scheme set. Represents the lens light field rays Light field rays with reference lens The difference in light between Represented in manifold space The square of the norm between x and y is the matching cost, where the light difference includes the light position difference and the light direction difference. Indicates the matching scheme The entropy value of represents the regularization coefficient that controls the matching complexity, Represents a matching scheme The x and y in the It means taking the minimum function value .

[0031] Among them, manifold space refers to a high-dimensional space, commonly used in geometry and mathematical physics. It is a nonlinear space that locally looks like Euclidean space, in which the points represent the position and direction of each light ray in the light field. In light field processing, manifold space is used to describe the relationship between light rays and how they change in different light fields. The concept of manifold is crucial to understanding how light propagates and changes in three-dimensional or higher-dimensional space. Specifically, the norm in manifold space measures the difference between light rays x and y in this space, including their spatial position and direction.

[0032] Optionally, the value range of λ can be adjusted between 0 and 1, such as 0.3 or 0.7.

[0033] It should be noted that, based on the Gromov-Wasserstein (GW) distance framework, the geometric structure difference between the lens light field and the reference light field is converted into a matching cost optimization problem. The optimal matching solution is solved by minimizing the matching cost (including position and direction differences) and the entropy regularization term between the two light fields. Among them, the entropy term controls the ambiguity of the matching: the smaller the entropy value, the clearer the matching solution. The reference light field is used as a sparse structure template to guide the algorithm to prioritize the retention of key light directions. Finally, by integrating the optimal transmission theory and entropy regularization, structure-aware compression decisions are achieved. While ensuring the structural similarity between the original light field and the reference light field, overly complex matching relationships are suppressed and redundant calculations are avoided, thereby directly reducing the search space and memory usage of the matching algorithm in the compression stage, reducing the amount of light field data, and laying a low-power foundation for the selective activation of subsequent photosensitive units through sparse mapping.

[0034] S203: Mapping the lens light field to the reference lens light field according to the optimal matching solution to compress the lens light field.

[0035] According to the optimal matching scheme, the original light field is mapped to the sparse manifold space of the reference light field. The geometric compression of the light field data is completed by eliminating redundant rays (such as rays that deviate greatly from the reference structure) and retaining core rays with low matching cost. This process not only reduces the amount of data, but also ensures that the geometric features of the compressed light field are consistent with the design goals.

[0036] It should be noted that by mapping the lens light field to the sparse manifold space of the reference light field, redundant light is eliminated and core light is retained, thereby effectively reducing the amount of data and the computational burden. At the same time, it ensures that the compressed light field is consistent with the design goals in terms of geometric features, which helps to optimize the power consumption and processing efficiency of the compact camera.

[0037] S3: Based on the compressed lens light field, the photosensitive unit activation matrix of the card camera is constructed based on the Keller geometry algorithm.

[0038] Among them, Kähler Geometry is an algorithm used in complex geometry, which is mainly used to describe and analyze geometric spaces with certain symmetries and structures. In image processing, Kähler Geometry can be used to construct a mathematical model of the light field, and to optimize the activation strategy of the photosensitive unit by studying the local changes and propagation of the light field. The photosensitive unit activation matrix is ​​a matrix that indicates whether the camera photosensitive unit needs to be activated, and each element represents the activation state of a photosensitive unit. The activation matrix constructed by the Kähler Geometry algorithm can accurately determine which photosensitive units need to be activated and which can remain closed based on the compressed lens light field, thereby reducing unnecessary power consumption and calculations.

[0039] It should be noted that the Keller geometry algorithm is used to analyze and process the compressed lens light field to construct a photosensitive unit activation matrix. This matrix accurately determines which photosensitive units need to be activated to capture effective image information and which can be turned off to reduce power consumption. Through this process, unnecessary calculations and power consumption can be significantly reduced while ensuring image quality, thereby optimizing the overall energy efficiency of the camera.

[0040] In a possible implementation, S3 specifically includes:

[0041] S301: Establishing propagation functions of the compressed lens light field at different positions on the photosensitive plane.

[0042] The specific propagation function is: ; in, represents pi, represents a natural constant, z represents a complex position point on the photosensitive plane, n represents the sub-beam number in the compressed lens light field, i represents a negative unit, represents the Gaussian envelope factor related to n, represents the negative exponential term describing the change of the light field related to i, n and z, Represents the propagation function value of the compressed lens light field at position z.

[0043] The propagation function is a superposition of complex waves, which is used to simulate the diffusion, interference and modulation behavior of light in space. It provides the light wave propagation information at each pixel position.

[0044] S302: According to the propagation function, the activation intensity potential energy field of the photosensitive units at different positions is constructed by using the Keller geometry algorithm.

[0045] The activation intensity potential field is calculated as follows: ; in, It represents the propagation function value of the compressed lens light field at position z, that is, the propagation intensity. represents the activation intensity potential field describing the local variation of the compressed lens light field at each position z, represents partial derivative, log represents logarithmic function, z0 represents visual focus, represents the complex conjugate of z0, Indicates taking z and The complex real part of the complex inner product between It represents the adjustment coefficient that controls the influence of visual focus on the activation intensity potential field. and Respectively represent z and Find the partial derivative, represents the complex conjugate of z.

[0046] Among them, Kähler geometry algorithm is a mathematical method that is widely used in complex geometric spaces, especially when studying geometric bodies with certain symmetries. In image processing, it is used to describe and calculate the propagation characteristics and local changes in the light field, and optimize the activation of photosensitive units through geometric principles. The activation intensity potential field is calculated based on the propagation function (light field propagation) and is used to describe the activation intensity of photosensitive units at different positions. This field helps determine the local changes of photosensitive units, that is, which photosensitive units should be activated by calculating the derivatives and geometric relationships of the propagation function. The field combines the influence of visual focus and adjusts the activation intensity at different positions through the regularization coefficient to optimize the image acquisition process. The activation intensity potential field of the photosensitive unit is accurately calculated through the Kähler geometry algorithm to ensure that only the required photosensitive units are activated, reducing unnecessary calculations and energy consumption, thereby significantly reducing power consumption while optimizing image quality and improving the overall energy efficiency of the camera.

[0047] Optionally, the adjustment coefficient is used to control the degree of influence of the visual focus on the activation intensity potential field. Specifically, the function is to adjust the influence of the complex inner product between the visual focus and other positions on the activation intensity, thereby determining which photosensitive units need to be activated. By changing the value of the adjustment coefficient, the importance of the focus in the activation of the photosensitive unit can be adjusted. For example, 0.3 or 0.6 can be taken.

[0048] S303: Convert the activation intensity potential energy field into a photosensitive unit activation matrix that describes whether each photosensitive unit is activated.

[0049] The photosensitive unit activation matrix is ​​specifically: ; in, Represents the photosensitive unit activation matrix.

[0050] It should be noted that by using the propagation function and Kähler geometry algorithm to accurately calculate the activation intensity potential field of the photosensitive unit, the refined control of the photosensitive unit is achieved. In this way, only those photosensitive units that contribute to the image quality can be activated to avoid unnecessary calculations and power consumption, thereby significantly reducing power consumption and improving the energy efficiency and performance of the camera while ensuring efficient image acquisition. This activation control method based on light field propagation and geometric optimization further improves the accuracy of power consumption optimization.

[0051] S4: Control the photosensitive unit of the card camera through the photosensitive unit activation matrix to collect real scene images.

[0052] It can be understood that by using the photosensitive unit activation matrix to control the photosensitive units of the card camera, it is ensured that only the required photosensitive units are activated to capture real-life images. Through this precise control, the activation of invalid photosensitive units is avoided, power consumption is reduced, and the collected image information is ensured to be complete and efficient. This method optimizes the performance of the camera and reduces unnecessary calculations and power consumption.

[0053] In a possible implementation manner, S4 is specifically: Determine whether the photosensitive unit activation element value in the photosensitive unit activation matrix is ​​greater than the preset photosensitive unit activation element value, if so, turn on the photosensitive unit corresponding to the photosensitive unit activation element value, otherwise, turn off the photosensitive unit corresponding to the photosensitive unit activation element value.

[0054] It should be noted that those skilled in the art can set the size of the preset photosensitive unit activation element value according to actual needs, and the present invention is not limited thereto.

[0055] It can be understood that, according to the comparison between the element value of the photosensitive unit activation matrix and the preset threshold value, if the element value is greater than the threshold value, the corresponding photosensitive unit is activated. Otherwise, the photosensitive unit is turned off. In this way, the switching of the photosensitive unit can be accurately controlled, the image acquisition process can be optimized, and unnecessary power consumption can be reduced.

[0056] S5: performing single conformal mapping on the real scene image to denoise the real scene image.

[0057] It should be noted that conformal mapping is a denoising method that preserves the geometric structure of an image. It adjusts the local geometric shape of the image to preserve the details and structure of the image while effectively removing unnecessary noise. Single conformal mapping preserves the characteristics of the image's geometric structure and converts the complex denoising process into the adjustment of local geometric shapes, avoiding redundant calculations of multi-step iterations or high-dimensional operations in traditional methods. This not only improves image quality, but also avoids excessive calculations and resource consumption, thereby optimizing power consumption control.

[0058] In a possible implementation, S5 specifically includes: S501: Construct the Weyl tensor of the real scene image.

[0059] Among them, the Weyl tensor is a mathematical tool for describing the change of curvature in space or images, and can be used for geometric analysis of bending or deformation. Specifically, the specific process of constructing the Weyl tensor of the real-life image is as follows: Calculation of local curvature of the image: First, calculate the curvature of the local geometric features of the image (such as edges, textures, etc.). Through derivative operations, the changes of each pixel in different directions can be obtained, and these changes reflect the local bending or deformation of the image. Then, differential geometry methods are used to analyze the curvature and deformation of the image, and the Weyl tensor helps quantify the geometric changes in different areas of the image. This process is mainly achieved by calculating the metric tensor and differential operator related to the local changes of the image. Then the perturbation information is extracted, that is, the main information captured by the Weyl tensor is the geometric perturbation in the image (such as local brightness changes and morphological deformation). This perturbation information helps to identify and suppress noise in the denoising process while retaining the structure and details of the image. Finally, a Weyl tensor describing the local geometric perturbation of the image is generated through mathematical formulas such as the Hessian matrix or the Riemann tensor. This tensor reflects the local curvature changes of the image and provides accurate geometric data for subsequent denoising and image optimization.

[0060] S502: With the goal of minimizing the difference between the Frobenius norm of the Weyl tensor and the peak signal-to-noise ratio of the real scene image, determine the denoising optimization objective function of the real scene image to obtain a Weyl tensor transformation operator that describes the local geometric disturbance characteristics of the real scene image.

[0061] The denoising optimization objective function is specifically: ; in, represents the Weyl tensor transformation operator that describes the local geometric perturbation characteristics of the real scene image. represents the Weyl tensor, express The square of the Frobenius norm, Represents a real-life image. Indicates passing The real scene image after mapping, represents the reference image, It means taking the minimum function value , express and The peak signal-to-noise ratio between Represents the regularization factor that adjusts the impact of peak signal-to-noise ratio.

[0062] Among them, the tensor transformation operator refers to the mathematical operation applied to the image, which describes how to geometrically transform or adjust the real-life image through the Weyl tensor to remove noise from the image. The reference image refers to an ideal or target image used as a comparison standard in the denoising process. In denoising optimization, the reference image is used to evaluate the quality of the denoised image, and the denoising effect is judged by calculating the peak signal-to-noise ratio (PSNR) of the two. The level of PSNR reflects the similarity between the denoised image and the reference image, thereby guiding the optimization process.

[0063] It should be noted that the denoising optimization objective function is constructed by minimizing the difference between the Frobenius norm of the Weyl tensor and the peak signal-to-noise ratio of the image. This optimization objective function takes into account the balance between the geometric perturbation of the image and the noise removal. Specifically, by minimizing the Frobenius norm of the Weyl tensor, the geometric distortion of the image can be reduced, while the image quality is ensured to be intact by adjusting the difference between the PSNR and the reference image. The core of this process is to find the optimal Weyl tensor transformation operator to remove noise through precise geometric adjustment while maintaining the details and structure of the image.

[0064] S503: Based on the acquired Weyl tensor transformation operator, a single conformal mapping is performed on the real scene image to denoise the real scene image.

[0065] Specifically, the advantage of this process is that by constructing a Weyl tensor and combining it with the denoising optimization objective function, the geometric disturbance characteristics of the image can be accurately controlled, thereby achieving efficient denoising. Specifically, the Weyl tensor transformation operator can effectively describe the local geometric changes of the image, ensuring that the image retains important details while denoising. Image quality is optimized by minimizing the difference between the Frobenius norm and the peak signal-to-noise ratio of the Weyl tensor. In addition, this process utilizes conformal mapping, which maintains the geometric structure of the image while simplifying the image transformation process. Specifically, conformal mapping is a transformation method that removes unnecessary complex geometric deformations and noise in the image by keeping the angle unchanged, reducing the amount of calculation and power consumption, avoiding excessive calculation of redundant information, improving the efficiency of power consumption optimization, and ensuring high-quality image output.

[0066] S6: Encode the denoised real scene image with the constraint of minimizing the entropy production value representing the loss of image information to obtain the target real scene image.

[0067] Among them, the entropy production value refers to the degree of information loss during the encoding process of the image. Entropy is a way to measure the amount of information. The smaller the entropy production value, the less information loss of the image. In image processing, minimizing the entropy production value means retaining as much useful information as possible when encoding the image and reducing the compression loss of the image. By constraining the encoding strategy to minimize the entropy production value, while ensuring image quality, the additional computing load caused by redundant data processing (such as repeated correction of high-frequency information or high-complexity compression) is reduced, thereby reducing the chip operation frequency and memory access times, directly reducing the power consumption of the encoding module, and avoiding subsequent correction operations caused by information loss (such as re-compression or frame supplementation), further optimizing the power consumption of compact cameras.

[0068] In a possible implementation, S6 specifically includes: S601: Calculate the image information entropy of the denoised real scene image.

[0069] S602: Establishing an evolution model of image information entropy in a metric space to simulate the diffusion process of image information of the denoised real scene image over time.

[0070] Among them, metric space is a mathematical concept, which means that the "distance" or "similarity" between elements in the space can be defined by a certain metric (distance function). In image processing, metric space is usually used to represent the structure and information content of an image, and the distance function can measure the difference between different image regions or pixels. Optionally, the metric space can be Euclidean space, Manhattan space, Hamming space or LP space.

[0071] The evolution model is used to describe the change process of image information. Specifically, this evolution model simulates the change of image information entropy over time through the diffusion equation, aiming to optimize the image denoising process through time evolution.

[0072] The specific evolution model is: ; in, represents partial derivative, t represents time variable, Represented in metric space The information entropy diffusion Laplace operator in , represents the gradient of information entropy H, Represented in metric space Next The square of the norm of Representing a metric space The Ricci curvature tensor of .

[0073] Among them, the information entropy diffusion Laplace operator is a Laplace operator defined in the metric space and applied to information entropy. It measures the diffusion and change rate of image information. In the context of image processing, the Laplace operator helps describe how information diffuses from one area to other areas. The diffusion process is related to the local structure and texture of the image and its relationship with the surrounding area. Using this operator can simulate how information propagates smoothly on the image and reduce noise. The Ricci curvature tensor is a tensor that describes the geometric properties of the manifold. It measures the impact of the local curvature of the manifold on the image structure in the metric space. Specifically, the Ricci curvature tensor reflects the degree of curvature of the image structure in different directions. By adding Ricci curvature, the evolution model can capture the differences in local geometric features of the image and further optimize the effect of image processing.

[0074] It should be noted that the advantage of this evolutionary model is that it comprehensively considers the geometric structure of the image and the information diffusion process. Through the influence of the diffusion Laplace operator and the Ricci curvature tensor of information entropy, it can accurately control the denoising process while retaining the structural information of the image. This method not only removes noise, but also enhances the propagation of useful information in the image, effectively improving the image quality.

[0075] S603: Solve the steady-state solution of the evolution model to obtain the optimal entropy distribution map of the real scene image under the minimum entropy production value.

[0076] S604: Encode the denoised real scene image based on the optimal entropy distribution map to obtain a target real scene image.

[0077] In a possible implementation, the encoding method in S604 specifically includes: lossy compression encoding and lossless compression encoding, wherein the lossy compression encoding includes JPEG and WebP, and the lossless compression encoding includes PNG, GIF, and TIFF.

[0078] In a possible implementation manner, S604 specifically includes:

[0079] S6041: Sort the information entropy of pixels at different positions in the denoised real scene image according to the optimal entropy distribution map.

[0080] S6042: Using the upper bit rate limit of the compact camera as a constraint, allocate bit rates in a manner inversely proportional to the information entropy value of the pixel points to encode the real scene image.

[0081] Specifically, firstly, the information entropy of pixels at different positions in the denoised image is sorted according to the optimal entropy distribution map to determine which areas contain more information. Then, under the bit rate upper limit constraint of the compact camera, the bit rate is allocated inversely proportionally according to the entropy value of the pixel points, that is, more bit rates (first preset bit rates) are allocated to areas with large amounts of information to ensure that important details are encoded with higher quality, while areas with less information use lower bit rates (second preset bit rates), thereby effectively compressing data while ensuring image quality, wherein the first preset bit rate is greater than the second preset bit rate. The specific bit rate can be adjusted according to actual needs.

[0082] Specifically, the denoising effect of an image can be optimized by calculating the information entropy of the image and using the evolution model in the metric space. First, the entropy value of the denoised image is calculated to measure the amount of information in the image. Then, a model of the evolution of information entropy over time is established to simulate the diffusion process of image information, and the diffusion equation is used to dynamically optimize the details and noise of the image. Next, the steady-state solution of the equation is solved to obtain the optimal image entropy distribution, ensuring that the information in the image is maximized and the irrelevant noise is minimized. Finally, the image is encoded according to the optimal entropy distribution map to obtain the target real-scene image, which ultimately optimizes the image quality and reduces the noise.

[0083] S7: Output the target real scene image.

[0084] In actual application, this scheme achieves high-efficiency imaging and energy consumption control through multi-stage collaborative processing. First, the four-dimensional light field data is captured by light field sensors, and the Gromov-Wasserstein distance is combined for structure-aware compression. The redundant light directions are eliminated by matching the optimal mapping of the light field manifold to reduce the amount of data. Subsequently, the photosensitive unit activation matrix is ​​constructed based on the Kähler geometry algorithm, and its symmetry and complex structure characteristics are used to accurately control the local response of the CMOS sensor, and only the photosensitive units in the key areas are activated to reduce circuit power consumption. In the image acquisition stage, dark current noise and computing load are reduced by dynamically shutting down unnecessary photosensitive units. Next, a single angle-preserving mapping is used to denoise the image, and high-frequency noise is eliminated by retaining the local angle relationship. Compared with the traditional filtering algorithm, the calculation iterations are reduced, and the camera power consumption is further optimized. Finally, the encoding optimization is performed with the entropy production value as the constraint condition, and the dynamic code length allocation is combined to suppress the block effect and frequency domain distortion, so that the compression process can reduce the memory access frequency while avoiding the heavy compression operation caused by information loss, and comprehensively reduce the computing frequency and power consumption of the card camera chip. The entire process achieves a balanced optimization of image quality and energy efficiency through the synergistic effect of light field compression, selective photosensitive activation, geometric shape-preserving denoising and entropy constrained coding.

[0085] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0086] In an embodiment of the present invention, first, the lens light field is obtained to provide complete optical information for subsequent image processing, ensuring the basic data quality of image processing. Then, the lens light field is compressed by the Gromov-Wasserstein distance, which effectively reduces redundant data, reduces power consumption from the source, and avoids the problem of sacrificing image quality to save power in traditional methods. Then, the Keller geometry algorithm is used to construct a photosensitive unit activation matrix, and only the required photosensitive units are activated, further reducing unnecessary calculations and power consumption. Then, the real-scene image captured by the card camera based on the photosensitive unit activation matrix is ​​denoised once through angle-preserving mapping, which can further save power consumption while reducing the impact of noise and improving image quality. Finally, an optimized encoding method based on entropy minimization is used to ensure that data is compressed without losing image quality, and finally a high-quality target image is output, effectively balancing power consumption and image quality. While significantly reducing power consumption, image quality is maintained.

[0087] Reference Manual Attached Figure 2 , showing a schematic structural diagram of a card camera power consumption optimization system based on image processing provided by the present invention.

[0088] The present invention further provides a card camera power consumption optimization system 20 based on image processing, which is applied to the above-mentioned card camera power consumption optimization method based on image processing, and comprises: Processor 201.

[0089] The memory 202 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 201 , the method for optimizing the power consumption of a compact camera based on image processing according to the method embodiment is implemented.

[0090] The card camera power consumption optimization system 20 based on image processing provided by the present invention can execute the above-mentioned card camera power consumption optimization method based on image processing and achieve the same or similar technical effects. To avoid repetition, the present invention will not elaborate on them.

[0091] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least: In an embodiment of the present invention, first, the lens light field is obtained to provide complete optical information for subsequent image processing, ensuring the basic data quality of image processing. Then, the lens light field is compressed by the Gromov-Wasserstein distance, which effectively reduces redundant data, reduces power consumption from the source, and avoids the problem of sacrificing image quality to save power in traditional methods. Then, the Keller geometry algorithm is used to construct a photosensitive unit activation matrix, and only the required photosensitive units are activated, further reducing unnecessary calculations and power consumption. Then, the real-scene image captured by the card camera based on the photosensitive unit activation matrix is ​​denoised once through angle-preserving mapping, which can further save power consumption while reducing the impact of noise and improving image quality. Finally, an optimized encoding method based on entropy minimization is used to ensure that data is compressed without losing image quality, and finally a high-quality target image is output, effectively balancing power consumption and image quality. While significantly reducing power consumption, image quality is maintained.

[0092] It should be understood that the processor in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0093] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0094] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When a computer instruction or computer program is loaded or executed on a computer, a process or function according to an embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0095] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0096] In the present invention, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0097] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0098] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0099] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0100] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0101] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0102] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0103] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0104] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, a method for optimizing power consumption of a compact camera based on image processing as described in a method embodiment is implemented.

[0105] A computer-readable storage medium provided by the present invention can implement the steps and effects of the card camera power consumption optimization method based on image processing in the above method embodiment. To avoid repetition, the present invention will not go into details.

[0106] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least: In an embodiment of the present invention, first, the lens light field is obtained to provide complete optical information for subsequent image processing, ensuring the basic data quality of image processing. Then, the lens light field is compressed by the Gromov-Wasserstein distance, which effectively reduces redundant data, reduces power consumption from the source, and avoids the problem of sacrificing image quality to save power in traditional methods. Then, the Keller geometry algorithm is used to construct a photosensitive unit activation matrix, and only the required photosensitive units are activated, further reducing unnecessary calculations and power consumption. Then, the real-scene image captured by the card camera based on the photosensitive unit activation matrix is ​​denoised once through angle-preserving mapping, which can further save power consumption while reducing the impact of noise and improving image quality. Finally, an optimized encoding method based on entropy minimization is used to ensure that data is compressed without losing image quality, and finally a high-quality target image is output, effectively balancing power consumption and image quality. While significantly reducing power consumption, image quality is maintained.

[0107] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0108] There are a few points to note:

[0109] (1) The drawings of the embodiments of the present invention only involve structures related to the embodiments of the present invention. Other structures may refer to conventional designs.

[0110] (2) For the sake of clarity, in the drawings used to describe the embodiments of the present invention, the thickness of layers or regions is exaggerated or reduced, that is, these drawings are not drawn according to the actual scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or there may be intermediate elements.

[0111] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0112] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A card camera power consumption optimization method based on image processing, characterized in that: include: S1: Obtain the lens light field of the card camera; S2: compressing the lens light field to reduce redundant data of the lens light field; S3: constructing a photosensitive unit activation matrix of the card camera based on the compressed lens light field and the Keller geometry algorithm; S4: controlling the photosensitive unit of the card camera through the photosensitive unit activation matrix to collect a real scene image; S5: performing single conformal mapping on the real scene image to denoise the real scene image; S6: Encode the denoised real scene image under the constraint of minimizing the entropy production value representing the loss of image information to obtain the target real scene image; S7: Output the target real scene image.

2. The card camera power consumption optimization method based on image processing according to claim 1, characterized in that: In step S2, the lens light field is compressed in combination with the Gromov-Wasserstein distance to reduce redundant data of the lens light field; The S2 specifically includes: S201: Obtaining geometric measurements, i.e., probability distributions, of light rays in the lens light field at different positions and directions; S202: establishing a matching relationship between the lens light field and a reference lens light field in combination with the geometric measurement, and outputting an optimal matching solution between the lens light field and the reference lens light field with the constraint of minimizing the matching cost between the lens light field and the reference lens light field; S203: Mapping the lens light field to the reference lens light field according to the optimal matching solution to compress the lens light field.

3. The card camera power consumption optimization method based on image processing according to claim 1, characterized in that: The S3 specifically includes: S301: Establishing a propagation function of the compressed lens light field at different positions of the photosensitive plane, wherein the propagation function is specifically: S302: constructing the activation intensity potential energy field of the photosensitive units at different positions according to the propagation function by using the Keller geometry algorithm; S303: Convert the activation intensity potential energy field into the photosensitive unit activation matrix that describes whether each photosensitive unit is activated.

4. The card camera power consumption optimization method based on image processing according to claim 1, characterized in that: The S4 is specifically: Determine whether the photosensitive unit activation element value in the photosensitive unit activation matrix is ​​greater than a preset photosensitive unit activation element value, if so, turn on the photosensitive unit corresponding to the photosensitive unit activation element value, otherwise, turn off the photosensitive unit corresponding to the photosensitive unit activation element value.

5. The card camera power consumption optimization method based on image processing according to claim 1, characterized in that: The S5 specifically includes: S501: Constructing the Weyl tensor of the real scene image; S502: determining a denoising optimization objective function of the real scene image with the goal of minimizing the difference between the Frobenius norm of the Weyl tensor and the peak signal-to-noise ratio of the real scene image, so as to obtain a Weyl tensor transformation operator that describes the local geometric disturbance characteristics of the real scene image; S503: Performing single conformal mapping on the real scene image based on the acquired Weyl tensor transformation operator to denoise the real scene image.

6. The card camera power consumption optimization method based on image processing according to claim 1, characterized in that: The S6 specifically includes: S601: Calculating the image information entropy of the denoised real scene image; S602: Establishing an evolution model of the image information entropy in a metric space to simulate the diffusion process of the image information of the denoised real scene image over time; S603: solving the steady-state solution of the evolution model to obtain an optimal entropy distribution diagram of the real scene image under the minimum entropy production value; S604: Encoding the denoised real scene image based on the optimal entropy distribution map to obtain the target real scene image.

7. The card camera power consumption optimization method based on image processing according to claim 6, characterized in that: The encoding method in S604 specifically includes: lossy compression encoding and lossless compression encoding, wherein the lossy compression encoding includes JPEG and WebP, and the lossless compression encoding includes PNG, GIF and TIFF.

8. The card camera power consumption optimization method based on image processing according to claim 6, characterized in that: The S604 specifically includes: S6041: sorting the information entropy of pixels at different positions in the denoised real scene image according to the optimal entropy distribution map; S6042: Using the bit rate upper limit of the card camera as a constraint, allocating the bit rate in a manner inversely proportional to the pixel information entropy value to encode the real scene image.

9. A card camera power consumption optimization system based on image processing, characterized in that: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method for optimizing power consumption of a compact camera based on image processing as claimed in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for optimizing power consumption of a card camera based on image processing as claimed in any one of claims 1 to 8 is implemented.

Images