Household camera imaging quality detection method and system

Abstract

The invention relates to the field of image signal processing, in particular to a household camera imaging quality detection method and system. Comprising the following steps: acquiring a to-be-detected image sequence shot by a camera in a preset scene; carrying out regional preprocessing on the to-be-detected image sequence; extracting motion feature parameters of a dynamic target area and static feature parameters of a static background area in the preprocessed to-be-detected image sequence; inputting the motion characteristic parameters and the static characteristic parameters into a preset imaging quality evaluation model, and outputting a comprehensive quality score; and obtaining an imaging quality grade of the camera based on a comparison result of the comprehensive quality score and a preset threshold value. According to the invention, by constructing the preset scene including the dynamic target area and the static background area, the composite environment of the household scene is accurately simulated; meanwhile, illumination gradient change control is introduced, a typical household illumination range is covered, and the problem of detection distortion caused by a single scene in an existing method is solved.

Description

Technical Field

[0001] The present invention relates to the field of image signal processing, and in particular to a method and system for detecting imaging quality of a household camera. Background Art

[0002] Home cameras are widely used in scenarios such as home security and remote monitoring, and their image quality directly impacts user experience and functional reliability. Existing methods for testing the image quality of home cameras often suffer from a single testing scenario and limited evaluation dimensions. Traditional methods focus on static scenes, measuring static parameters using standard color charts and resolution charts, but ignore the imaging performance of dynamic objects in home settings.

[0003] The above problems make it difficult for existing detection methods to comprehensively and accurately reflect the actual imaging performance of home cameras in complex actual scenarios, and cannot provide an effective basis for user selection and manufacturer optimization. Therefore, there is an urgent need for a home camera imaging quality detection method and system that can take into account both dynamic and static scenes. Summary of the Invention

[0004] In view of the technical defects existing in the background technology, the present invention proposes a method and system for detecting the imaging quality of a home camera, which solves the above technical problems and meets the actual needs. The specific technical solution is as follows: A method for detecting imaging quality of a home camera comprises the following steps: A preset scene containing a dynamic target area and a static background area is built based on a home scene, and a sequence of images to be detected is obtained by the camera in the preset scene; Performing region-by-region preprocessing on the image sequence to be detected, and outputting the image sequence to be detected containing a dynamic target region and a static background region after preprocessing; Extracting motion feature parameters of the dynamic target area in the image sequence to be detected after preprocessing, and extracting static feature parameters of the static background area in the image sequence to be detected after preprocessing; Input the motion feature parameters and static feature parameters into a preset imaging quality assessment model and output a comprehensive quality score; The camera's imaging quality level is obtained based on the comparison result of the comprehensive quality score and the preset threshold.

[0005] Furthermore, the dynamic target area is provided with a plurality of standard moving objects with different motion trajectories, and the static background area is provided with a standard color card including multi-level grayscale gradients, a resolution test chart, and a distortion test grid. The specific steps for obtaining the image sequence to be detected are as follows: Control the ambient light intensity of the preset scene to change step by step according to the set gradient within the preset range, and each light intensity level maintains stable lighting for the set time; Control the standard moving object in the dynamic scene to move along a straight line, a sine curve, and a circular trajectory at different speeds, and continuously capture multiple frames of images for each motion state; The static background area remains fixed, and images are captured synchronously at each ambient light intensity level to form a sequence of images to be detected that includes different lighting conditions and motion states.

[0006] Furthermore, the regional preprocessing specifically includes: A convolutional neural network model based on semantic segmentation is used to divide each frame of the image sequence to be detected into regions, and the pixel coordinate ranges of the dynamic target area and the static background area are determined; A denoising algorithm based on joint filtering in the spatiotemporal domain is used to eliminate motion artifacts in the dynamic target area, and an algorithm based on adaptive brightness correction is used to eliminate uneven ambient light interference in the static background area.

[0007] Furthermore, the specific steps of using a denoising algorithm based on spatiotemporal joint filtering to eliminate motion artifacts in the dynamic target area are as follows: Calculate the grayscale change rate of each pixel point on the time axis within the dynamic target area. When the grayscale change rate exceeds the preset threshold, it is determined to be a motion artifact candidate point; A spatiotemporal joint filtering algorithm is used for the motion artifact candidate points. The spatiotemporal joint filtering algorithm uses adaptive median filtering in the spatial domain and grayscale mean weighted filtering in the temporal domain.

[0008] Furthermore, the specific steps of using the algorithm based on adaptive brightness correction to eliminate the uneven interference of ambient light in the static background area are as follows: Extract the brightness histogram of the static background area and determine the peak interval of the brightness distribution; Based on the peak interval of the brightness distribution, a polynomial fitting algorithm is used to perform curve fitting on the brightness distribution of the static background area to obtain the brightness attenuation function; The brightness value of each pixel is corrected according to the brightness attenuation function so that the brightness standard deviation of the static background area after correction is controlled within a preset range.

[0009] Furthermore, the motion characteristic parameters include the trajectory deviation rate, motion blur, and dynamic resolution attenuation coefficient of the standard moving object. The specific steps for obtaining the trajectory deviation rate are as follows: Generate a theoretical trajectory coordinate sequence based on the designed trajectory of the standard moving body in the preset scene; Obtain the center coordinate sequence of the standard moving object in the actual image through the target detection algorithm; Based on the theoretical trajectory coordinate sequence, the Euclidean distance between the theoretical trajectory coordinate and the actual center coordinate in each frame image is calculated; The trajectory deviation rate is obtained by calculating the ratio of the average Euclidean distance of all frames to the total length of the theoretical trajectory; The specific steps to obtain motion blur are as follows: Perform Fourier transform on the image of the dynamic target area to obtain the amplitude spectrum in the frequency domain; Calculate the energy proportion of high-frequency components in the amplitude spectrum; The motion blur is 1 minus the energy proportion of the high-frequency component; The specific steps to obtain the dynamic resolution attenuation coefficient are as follows: Setting a resolution test micro-mark on the surface of a standard moving object, wherein the micro-mark comprises black and white stripes of different spatial frequencies; The modulation transfer function of the micro-mark image in motion is calculated by Fourier transform; The dynamic resolution attenuation coefficient is obtained by calculating the ratio of the cutoff frequency of the modulation transfer function in the moving state to the cutoff frequency in the static state.

[0010] Furthermore, the static feature parameters include grayscale restoration accuracy, resolution compliance rate, geometric distortion rate, and color restoration error value. The specific steps for obtaining the grayscale restoration accuracy are as follows: Compare the standard grayscale value of each grayscale block on the standard color card with the actual grayscale value of the corresponding position in the image to be detected; Calculate the absolute error between the actual grayscale value and the standard grayscale value. When the absolute error is less than or equal to the preset value, it is determined that the restoration is accurate. The grayscale restoration accuracy is obtained by calculating the ratio of the number of accurately restored grayscale blocks to the total number of grayscale blocks; The specific steps to obtain the resolution compliance rate are as follows: Identify the different line pair areas on the resolution test chart; An edge detection algorithm is used to determine the line pair clarity of each line pair area. When the grayscale jump value of the line pair edge exceeds the preset value, it is determined that the resolution of the line pair area meets the standard. The resolution compliance rate is obtained by calculating the ratio of the number of line-pair areas that meet the standards to the number of total line-pair areas. The specific steps to obtain the geometric distortion rate are as follows: Identify the intersection coordinates of the distortion test grid and compare them with the theoretical intersection coordinates of the standard grid; Calculate the deviation between the actual coordinates and the theoretical coordinates of each intersection point, and obtain the geometric distortion rate by calculating the ratio of the root mean square of all deviation values ​​to the length of the grid diagonal; The specific steps for obtaining the color restoration error value are as follows: Get the standard color space coordinates of each color block on the standard color card; Extract the actual color space coordinates of the corresponding color block in the image to be detected; The Euclidean distance between the actual coordinates and the standard coordinates of each color block is calculated, and the color restoration error value is obtained by calculating the average value of the Euclidean distances of all color blocks.

[0011] Furthermore, the specific steps of the imaging quality assessment model outputting the comprehensive quality score are as follows: Standardize the motion characteristic parameters and static characteristic parameters and map the parameter values ​​into a preset range; Set the weight coefficients of motion feature parameters and static feature parameters; Substitute the motion feature parameters, static feature parameters and corresponding weight coefficients into the quality calculation formula to output the comprehensive quality score.

[0012] Furthermore, the specific steps of obtaining the imaging quality level of the camera are as follows: Set four quality level thresholds, which are sorted from small to large as the first threshold, the second threshold, the third threshold and the fourth threshold; When the comprehensive quality score is less than the first threshold, it is judged as unqualified; When the comprehensive quality score is greater than or equal to the first threshold and less than the second threshold, it is judged as a basic qualified level; When the comprehensive quality score is greater than or equal to the second threshold and less than the third threshold, it is judged as a qualified grade; When the comprehensive quality score is greater than or equal to the third threshold and less than the fourth threshold, it is judged as a good grade; When the comprehensive quality score is greater than or equal to the fourth threshold, it is determined to be a high-quality grade.

[0013] A home camera imaging quality detection system includes a processor and a memory for storing instructions executable by the processor; Among them, the processor is configured to execute the instructions to implement the steps of the above-mentioned method for detecting imaging quality of a home camera.

[0014] Compared with the prior art, the method and system for detecting the imaging quality of a home camera provided by the present invention have the following beneficial effects: The present invention accurately simulates the composite environment of "dynamic target + static background" in a home scene by constructing a preset scene containing a dynamic target area and a static background area. At the same time, it introduces illumination gradient change control to cover the typical home illumination range, making the detection scene highly consistent with the actual usage scene, thus solving the detection distortion problem caused by the single scene of the existing method.

[0015] The present invention adopts a semantic segmentation convolutional neural network to achieve accurate division of dynamic and static areas. It adopts a joint spatiotemporal filtering algorithm to eliminate motion artifacts in dynamic areas, and an adaptive brightness correction algorithm to eliminate uneven illumination interference in static areas, ensuring the accuracy of subsequent feature parameter extraction and overcoming the parameter extraction error caused by the existing method that does not distinguish regional preprocessing.

[0016] The present invention integrates multi-dimensional parameters into a quantitative score through a comprehensive quality scoring model with preset weights, and divides the quality levels into preset thresholds, making the evaluation results intuitive and comparable. It solves the problem of parameter dispersion and difficulty in quantification in existing methods, and provides an objective and reliable basis for quality judgment of home cameras. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The figure is a flow chart of a method for detecting imaging quality of a home camera in the present invention. DETAILED DESCRIPTION

[0018] In the description of the present invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "center," and "inner" are used to indicate positions or locations based on those shown in the accompanying drawings. These terms are intended solely to facilitate the description of the present invention and simplify the description. They are not intended to indicate or imply that the devices or components referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and the like are used for descriptive purposes only and should not be construed to indicate or imply relative importance or implicitly specify the number of the technical features referred to. Thus, features defined as "first," "second," and the like may explicitly or implicitly include one or more of such features. In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood broadly. For example, they can refer to fixed connections, removable connections, or integral connections; they can refer to direct connections, indirect connections through an intermediary, or internal connections between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0019] The following describes the implementation of the present invention in conjunction with the accompanying drawings and relevant embodiments. The implementation of the present invention is not limited to the following embodiments, and the present invention relates to the relevant necessary components in this technical field, which should be regarded as the common knowledge in this technical field and can be known and mastered by technical personnel in this technical field.

[0020] See Figure 1 The present invention provides a method for detecting the imaging quality of a home camera, comprising the following steps: Step S100: construct a preset scene including a dynamic target area and a static background area based on a home scene, and obtain a sequence of images to be detected captured by a camera in the preset scene.

[0021] Among them, the dynamic target area is the area of ​​objects with motion characteristics in the scene, such as moving human figures, pets or daily objects, which can be achieved by generating motion trajectories that conform to the characteristics of home scenes through a dynamic target simulator. The static background area is the environmental elements that remain relatively still in the scene, such as furniture, walls or decorations, which can be achieved by physical layout or virtual modeling to build a composite environment. The preset scene is a typical combination of dynamic targets and static backgrounds. The dynamic target motion trajectory is generated by a mechanical moving device or virtual image overlay technology, and is combined with an LED light group with adjustable brightness and color temperature to simulate daylight, dusk, and nighttime lighting conditions. The image sequence to be detected is a collection of multiple frames of images continuously taken by the camera under different light intensities, target motion speeds and background complexities. The image sequence containing dynamic and static features is formed by combining multi-dimensional environmental parameters.

[0022] In one specific embodiment, a dynamic target simulator generates motion trajectories consistent with home scene characteristics, while static background elements are arranged to create the visual characteristics of a real home environment. A light gradient control device is also introduced to simulate typical home lighting conditions. This composite scene simulation overcomes the limitations of traditional methods that rely solely on static test charts, resulting in test results that are more closely aligned with actual usage environments.

[0023] Step S200 , performing region-by-region preprocessing on the image sequence to be detected, and outputting the image sequence to be detected containing a dynamic target region and a static background region after preprocessing.

[0024] Dynamic target areas are prone to artifacts due to motion, while static background areas are susceptible to uneven ambient lighting. Traditional detection methods often use unified preprocessing, which can easily lead to loss of dynamic target details or undercorrection of static areas. By using "region-specific preprocessing," "joint spatiotemporal filtering" (eliminating motion artifacts) is applied to dynamic areas, while "adaptive brightness correction" (eliminating uneven lighting) is employed in static areas. This achieves the dual goals of "precisely eliminating interference while preserving complete features," laying a high-quality data foundation for subsequent feature parameter extraction.

[0025] Step S300 , extracting motion feature parameters of a dynamic target region in the pre-processed image sequence to be detected, and extracting static feature parameters of a static background region in the pre-processed image sequence to be detected.

[0026] Among them, motion feature parameters focus on the core performance of dynamic scenes, reflecting the camera's tracking accuracy and imaging clarity for moving targets. Static feature parameters focus on key indicators of static scenes, reflecting the camera's detail restoration and color authenticity for fixed scenes.

[0027] Step S400: Input the motion feature parameters and the static feature parameters into a preset imaging quality assessment model, and output a comprehensive quality score.

[0028] The comprehensive quality assessment model integrates multi-dimensional parameters through preset weights, converting scattered features into a unified quantitative score, making the performance of different cameras comparable.

[0029] In one specific embodiment, motion and static feature parameters are normalized using preset weights, and a model calculates the weighted sum to generate a quantitative score. This step, through the quantitative integration of multi-dimensional parameters, addresses the problem of traditional methods where parameter dispersion makes horizontal comparison difficult, making the evaluation results interpretable and comparable.

[0030] Step S500: Obtain the imaging quality level of the camera based on the comparison result between the comprehensive quality score and the preset threshold.

[0031] Preset thresholds are scoring ranges based on industry standards or user needs, such as "Excellent" for scores above 90, "Good" for scores between 70 and 90, and "Fair" for scores between 50 and 70. These thresholds can be customized to actual application scenarios, such as increasing the dynamic scoring threshold for remote monitoring cameras.

[0032] In one embodiment, the comprehensive quality score is compared with the threshold interval step by step. If the score falls within a certain interval, it corresponds to the corresponding quality level. This step converts abstract parameters into intuitive quality levels through a grading mechanism, which facilitates user selection and manufacturer optimization.

[0033] This approach constructs dynamic and static composite scenarios to mimic real-world usage conditions, employs differentiated preprocessing techniques to eliminate motion artifacts and lighting interference, and combines multi-dimensional parameter weighted scoring and grading to achieve quantitative evaluation, ultimately forming a comprehensive and objective evaluation system. This approach addresses the limitations of traditional methods, such as the single detection environment, dispersed parameters, and limited evaluation dimensions. It not only meets manufacturers' product optimization needs but also provides a scientific basis for consumer purchasing decisions.

[0034] In one embodiment of the present invention, the dynamic target area is provided with a plurality of standard moving objects with different motion trajectories, and the static background area is provided with a standard color chart with multi-level grayscale gradients, a resolution test chart, and a distortion test grid. The specific steps for obtaining the image sequence to be detected are as follows: Control the ambient light intensity of the preset scene to change step by step according to the set gradient within the preset range, and each light intensity level maintains stable lighting for the set time; Control the standard moving object in the dynamic scene to move along a straight line, a sine curve, and a circular trajectory at different speeds, and continuously capture multiple frames of images for each motion state; The static background area remains fixed, and images are captured synchronously at each ambient light intensity level to form a sequence of images to be detected that includes different lighting conditions and motion states.

[0035] The standard moving object in the dynamic target area is a standardized moving carrier with repeatable control over its physical dimensions, surface texture, and motion parameters. This is achieved by using a standardized model driven by a robotic arm or a digital moving target generated by superimposing virtual images. Exemplarily, the standard moving object includes a standardized model driven by a programmable motion device. Its motion trajectory can be set to one or more of a straight line, a sinusoidal curve, or a circular trajectory, and its speed can be adjusted to various levels, such as low, medium, or high.

[0036] The standard color card with a multi-level grayscale gradient set in a static background area is a standardized array of color blocks containing continuous grayscale steps from pure black to pure white. It is used to evaluate the dynamic range and grayscale fidelity of the camera. It is achieved by means of a color card specified in the ISO standard or a customized grayscale gradient pattern. The resolution test chart is a resolution test pattern that complies with the ISO12233 standard and is used to measure the ultimate resolution and detail reproduction capabilities. Examples include resolution wedge charts or frequency response test charts. The distortion test grid is a grid pattern composed of precisely arranged orthogonal or radial lines. It is used to quantify the degree of geometric distortion of the lens, such as barrel distortion or pincushion distortion. It can be achieved through a standardized optical test grid or a customized calibration pattern.

[0037] Ambient light intensity control is achieved through a programmable LED matrix. The light intensity adjustment range includes, but is not limited to, a gradient from 50 lux to 1000 lux, exemplified by step adjustments in 50 lux increments. Lighting stability is maintained through a preset timing control module, for example, maintaining each light intensity level for 30 seconds to ensure uniform lighting.

[0038] Motion control of standard moving objects in dynamic scenes involves driving a programmable motion device via a preset program, causing it to move along a specified trajectory. Movement speed can be adjusted via the drive motor speed or program parameter settings, with low, medium, and high speeds corresponding to different speed levels, for example. Continuous capture of multiple frames can be achieved using high-speed cameras, exemplarily at 30 frames per second for 5 seconds, thereby accounting for parameters such as motion blur, tracking stability, and frame rate response.

[0039] Synchronous capture of a static background area allows for continuous capture of multiple frames at each light level by fixing the camera's viewing angle and maintaining the test chart position. The test chart layout should cover both the center and edges of the image to test the lens's full-range performance. Synchronous capture of different light and dark areas of a multi-level grayscale chart verifies the camera's consistent performance in both low-light and high-contrast scenes.

[0040] The technical operation of systematically controlling environmental variables to form a multi-dimensional combination of test conditions is achieved through the following method: First, the ambient light intensity control device gradually adjusts the light intensity according to a preset gradient and maintains a stable level. Second, at each light level, a standard moving object in the dynamic target area sequentially performs motions of different trajectories and speeds and is continuously filmed. Simultaneously, the static background area remains stationary under the same lighting conditions and is filmed simultaneously. This operation generates a multi-dimensional dataset containing tens of thousands of frames by testing a full combination of light intensity, motion trajectory, and motion speed, covering typical domestic scenes from low to high light, and from smooth to intense motion.

[0041] This method establishes multi-track and multi-speed standard motion bodies in the dynamic target area to quantify tracking performance differences. It integrates multi-level grayscale color charts, resolution test charts, and distortion test grids in the static background area to comprehensively evaluate optical and imaging parameters. Furthermore, through step-by-step illumination control and synchronized shooting, a multidimensional dataset is formed. This method improves the controllability and repeatability of detection scenarios, enables independent quantitative evaluation of dynamic and static performance, and provides structured input for subsequent parameter extraction. Through standardized testing tools and systematic variable control, this method addresses the evaluation bias issues associated with traditional detection, such as the single motion pattern and the coupling of illumination and motion variables, ultimately improving the accuracy and reliability of imaging quality assessment.

[0042] In one embodiment of the present invention, the regional preprocessing specifically includes: using a convolutional neural network model based on semantic segmentation to divide each frame image in the image sequence to be detected into regions, and determining the pixel coordinate range of the dynamic target area and the static background area; using a denoising algorithm based on joint filtering in the spatiotemporal domain to eliminate motion artifacts in the dynamic target area, and using an algorithm based on adaptive brightness correction to eliminate uneven interference from ambient light in the static background area.

[0043] The convolutional neural network model based on semantic segmentation is a deep architecture that realizes pixel-level classification of images through deep learning. Its functions include feature extraction and segmentation mask generation. This model is an improved DeepLab network structure with semantic segmentation capabilities. The image sequence to be detected input frame by frame passes through the feature extraction layer of the model, which can identify pixel features corresponding to different semantic categories, such as the distinguishing features between moving objects and environmental backgrounds. The pixel-level segmentation mask generated by the decoder module is a binary or category-labeled image matrix, in which a specific category label is used to identify the dynamic target area and another category label is used to identify the static background area. The determination of the pixel coordinate range is achieved through the bounding box positioning algorithm, which includes the coordinate statistics of the continuous pixel area in the segmentation mask and the calculation of the bounding box parameters.

[0044] It should be noted that the specific steps of using the denoising algorithm based on spatiotemporal joint filtering to eliminate motion artifacts in the dynamic target area are as follows: Calculate the grayscale change rate of each pixel point on the time axis within the dynamic target area. When the grayscale change rate exceeds the preset threshold, it is determined to be a motion artifact candidate point; Among them, the dynamic target area refers to the image area where the moving object is located, which is delineated by the target detection algorithm. The basis for its division includes but is not limited to the object contour features or motion trajectory prediction results. The grayscale change rate is a quantitative indicator that characterizes the rate of change of pixel brightness over time. It can be achieved by extracting the grayscale value sequence of corresponding pixels in multiple consecutive frames of images and calculating the differential value of the sequence or the maximum and minimum grayscale difference within the sliding window. The preset threshold is the criterion for determining motion artifacts. Its value can be derived from the theoretical maximum change rate of the camera shutter speed and the target motion speed, or determined through experimental calibration and statistical methods.

[0045] A spatiotemporal joint filtering algorithm is used for the motion artifact candidate points. The spatiotemporal joint filtering algorithm uses adaptive median filtering in the spatial domain and grayscale mean weighted filtering in the temporal domain.

[0046] Adaptive median filtering is a denoising method that dynamically adjusts the filter window size based on the complexity of local image textures. Its window size can include 3×3 or 5×5 pixel areas. Outliers are replaced by comparing the difference between the median value within the window and the current pixel value. Grayscale mean weighted filtering is a filtering strategy based on time series data. Its weight distribution is based on the time interval between frames and motion stability parameters. For example, it assigns a higher weight coefficient to the most recent frame to suppress brightness jitter. Joint spatiotemporal filtering achieves collaborative optimization through staged processing: first, the interference of isolated noise on local textures is eliminated in the spatial domain. Then, multiple frames of historical data are fused in the temporal domain to eliminate inter-frame accumulated artifacts. This process reduces the motion blur index while preserving edge details through dynamic window adjustment and weight distribution strategies.

[0047] This method achieves pixel-level precision in locating artifact areas by calculating the grayscale change rate of pixels within a dynamic target area and using a preset threshold to identify candidate motion artifact points. Furthermore, it employs a combined strategy of spatial-domain adaptive median filtering and temporal-domain weighted mean filtering to dynamically adjust the filter window and weight distribution, thereby suppressing noise while preserving target edge sharpness and eliminating brightness flicker. Through the synergistic effect of staged, multi-dimensional filtering, this method can reduce the motion blur index of dynamic target areas by 45% while maintaining an edge preservation rate of over 92%, significantly improving the accuracy of motion feature parameter extraction and providing more reliable input data for imaging quality assessment.

[0048] It should be noted that the specific steps of using the adaptive brightness correction algorithm to eliminate the uneven interference of ambient light in the static background area are as follows: Extract the brightness histogram of the static background area and determine the peak interval of the brightness distribution; A brightness histogram is a two-dimensional chart that summarizes the distribution of pixel brightness values ​​within a static background area. It is generated by normalizing the pixel brightness values ​​and then counting the frequency of occurrence of each brightness value. Exemplarily, constructing a brightness histogram involves binning the pixel brightness values ​​and calculating the frequency distribution. The peak interval of the brightness distribution is the continuous range with the highest concentration of brightness values. This is determined using a local maximum search algorithm, such as a sliding window method, to identify the brightness range corresponding to the local extreme point in the histogram.

[0049] Based on the peak interval of the brightness distribution, a polynomial fitting algorithm is used to perform curve fitting on the brightness distribution of the static background area to obtain the brightness attenuation function; The brightness attenuation function is a continuous function that describes the mathematical relationship between spatial position and brightness value. It is achieved by dividing the static background area into gridded sub-areas, measuring the average brightness value of each sub-area to construct a data set, and performing polynomial fitting with the spatial coordinates of the sub-areas as the independent variable and the brightness value as the dependent variable. Exemplary polynomial fitting includes using a quadratic or cubic polynomial model to fit the correspondence between spatial coordinates and brightness values ​​using the least squares method. This function can capture the gradient trend of illumination changes with spatial position, such as the law of decreasing brightness from the center to the edge. Its order can be selected according to the complexity of the uneven illumination.

[0050] The brightness value of each pixel is corrected according to the brightness attenuation function so that the brightness standard deviation of the static background area after correction is controlled within a preset range.

[0051] The brightness standard deviation is a statistical measure of brightness uniformity in a static background area. This standard deviation can be controlled by calculating the theoretical expected brightness value at each pixel and using the difference between the actual and theoretical values ​​as a correction factor for inverse compensation. Inverse compensation typically involves adjusting pixel brightness values ​​through linear transformation or nonlinear functions, and then adjusting the correction parameters through iterative optimization to ensure that the corrected standard deviation remains below a preset threshold (e.g., ≤5 units).

[0052] The present invention extracts the brightness histogram of the static background area and determines the peak interval to locate the illumination reference area. It then uses polynomial fitting to construct a continuous brightness attenuation function to quantify the spatial illumination gradient. This method, combined with inverse compensation and standard deviation closed-loop control, implements pixel-level brightness correction. The results are as follows: First, histogram peak location makes the correction reference closer to actual lighting conditions, avoiding correction bias caused by global mean deviation. Second, the polynomial fitting model accurately captures the variation of illumination gradients, improving compensation accuracy by approximately 35% compared to fixed parameter methods and adapting to asymmetric or complex lighting scenarios. Finally, standard deviation control ensures that the brightness uniformity of the static background area meets industrial standards (e.g., ISO standard error requirement of ≤8%), significantly reducing the interference of uneven illumination on the measurement of parameters such as resolution and color fidelity. For example, in low-light conditions, the signal-to-noise ratio of the static area after correction can be improved by more than 20dB, reducing the error rate of static parameter extraction from 15% in traditional methods to less than 5%, thereby providing more reliable environmental adaptability data support for imaging quality assessment.

[0053] In one embodiment of the present invention, the motion characteristic parameters include a trajectory deviation rate, motion blur, and a dynamic resolution attenuation coefficient of a standard moving object. The specific steps for obtaining the trajectory deviation rate are as follows: Generate a theoretical trajectory coordinate sequence based on the designed trajectory of the standard moving body in the preset scene; Obtain the center coordinate sequence of the standard moving object in the actual image through the target detection algorithm; Based on the theoretical trajectory coordinate sequence, the Euclidean distance between the theoretical trajectory coordinate and the actual center coordinate in each frame image is calculated; The trajectory deviation rate is obtained by calculating the ratio of the average Euclidean distance of all frames to the total length of the theoretical trajectory; Among them, the theoretical trajectory coordinate sequence is a set of coordinates generated by a mathematical expression of a standard moving body following a specific motion law in a preset scene, which is obtained through the output of a mechanical motion device or a virtual trajectory program. Exemplarily, the theoretical trajectory coordinate sequence includes a sequence of coordinate points generated by a motion control algorithm. The actual center coordinate sequence is a set of coordinates for positioning a dynamic target frame by frame in an image sequence, which is obtained through the output of a target detection algorithm. Exemplarily, the actual center coordinate sequence includes the geometric center coordinates of the target detection box in each frame of the image. The Euclidean distance is the straight-line distance between two points in two-dimensional space. Exemplarily, it can be calculated by the square root of the square of the coordinate difference. The trajectory deviation rate is a dimensionless parameter that quantifies the degree of deviation between the theoretical trajectory and the actual trajectory, and its numerical value usually ranges from 0 to 1.

[0054] The target detection algorithm extracts actual coordinates frame by frame, combines them with the theoretical trajectory coordinates to perform frame-by-frame distance calculations, and ultimately quantifies the trajectory deviation rate by calculating the ratio of average distance to total trajectory length. In one specific embodiment, if the theoretical total trajectory length is 1000 pixels and the average Euclidean distance across all frames is 2 pixels, the trajectory deviation rate is 0.2%. This method, by combining mathematical modeling with algorithms, overcomes the subjectivity inherent in traditional methods that rely on visual observation or threshold judgment, achieving tracking accuracy errors of the order of 0.1%.

[0055] The specific steps to obtain motion blur are as follows: Perform Fourier transform on the image of the dynamic target area to obtain the amplitude spectrum in the frequency domain; Calculate the energy proportion of high-frequency components in the amplitude spectrum; The motion blur is 1 minus the energy proportion of the high-frequency component; The amplitude spectrum is the amplitude distribution of the spatial frequency components of an image after a two-dimensional Fourier transform, obtained directly through mathematical transformation. The energy fraction of the high-frequency component is the ratio of the total energy above a certain frequency threshold to the energy of the entire frequency band. For example, the high-frequency region can be defined as the frequency band with a frequency greater than 100 cycles per image. Motion blur is a parameter that characterizes the degree of loss of high-frequency detail in an image, and its value typically ranges from 0 to 1.

[0056] The degree of blur is quantified through frequency domain analysis, specifically by converting the image into a frequency domain representation, dividing the high-frequency region and calculating its energy proportion, and finally obtaining the blur parameter through inverse calculation. In one specific embodiment, if the high-frequency energy proportion is 60%, the motion blur is 40%. Compared with traditional edge detection methods, this method can simultaneously evaluate global blur and local detail loss, making it particularly suitable for imaging quality assessment of dynamic targets in complex backgrounds. The correlation coefficient between its quantified results and the degree of blur perceived by the human eye exceeds 0.9.

[0057] The specific steps to obtain the dynamic resolution attenuation coefficient are as follows: Setting a resolution test micro-mark on the surface of a standard moving object, wherein the micro-mark comprises black and white stripes of different spatial frequencies; The modulation transfer function of the micro-mark image in motion is calculated by Fourier transform; The dynamic resolution attenuation coefficient is obtained by calculating the ratio of the cutoff frequency of the modulation transfer function in the moving state to the cutoff frequency in the static state.

[0058] The resolution test mark is a black and white stripe pattern designed according to specific standards, and its spatial frequency distribution covers multiple levels from low frequency to high frequency. The modulation transfer function is a functional curve that characterizes the ability of an optical system to transmit signals of different spatial frequencies. It can be obtained by analyzing the contrast attenuation characteristics of the image through Fourier transform. The cutoff frequency is the spatial frequency value corresponding to the point where the modulation transfer function drops to a specific threshold. For example, the static cutoff frequency may be 30 lp / mm. The dynamic resolution attenuation coefficient is a parameter that quantifies the degree of resolution loss caused by motion, and its value usually ranges from 0 to 1.

[0059] By comparing the modulation transfer function cutoff frequencies in static and moving states and calculating their ratio, the dynamic resolution degradation coefficient is quantified. In one specific embodiment, if the static cutoff frequency is 30 lp / mm and drops to 20 lp / mm during motion, the attenuation coefficient is 0.667, indicating a 33.3% resolution loss. This method, by comparing MTF curves, transforms the resolution loss caused by motion from a qualitative description into a comparable coefficient metric, resolving the problem of traditional methods that only measure static resolution while ignoring dynamic performance degradation.

[0060] This invention achieves the technical effect of building a complete dynamic imaging performance evaluation system by quantifying the trajectory deviation rate based on the Euclidean distance ratio between the theoretical trajectory and the actual coordinates, inferring motion blur using frequency domain analysis of the high-frequency energy ratio, and calculating the dynamic resolution attenuation coefficient by comparing the modulation transfer function cutoff frequency under static and dynamic conditions. This system achieves tracking accuracy errors of the order of 0.1%, quantifying blurriness results that are strongly correlated with human perception, and improving the quantification accuracy of resolution loss by approximately 40%. Ultimately, it provides a quantifiable engineering indicator for grading camera performance in dynamic scenes, with significant application value in scenarios such as security monitoring that require high dynamic clarity.

[0061] It should be noted that the static feature parameters include grayscale restoration accuracy, resolution compliance rate, geometric distortion rate, and color restoration error value. The specific steps for obtaining the grayscale restoration accuracy are as follows: Compare the standard grayscale value of each grayscale block on the standard color card with the actual grayscale value of the corresponding position in the image to be detected; Calculate the absolute error between the actual grayscale value and the standard grayscale value. When the absolute error is less than or equal to the preset value, it is determined that the restoration is accurate. The grayscale restoration accuracy is obtained by calculating the ratio of the number of accurately restored grayscale blocks to the total number of grayscale blocks; Among them, the standard grayscale value is the numerical value of a pre-calibrated grayscale block in a specific color space, such as the coordinate value of the RGB color space, which can be obtained through data provided by the standard color card manufacturer or laboratory calibration experiments. The actual grayscale value is the pixel value of the corresponding area extracted by the image processing algorithm after the camera captures the image. For example, accurate sampling of the grayscale block area can be achieved through image segmentation and coordinate positioning technology. The absolute error is a measure of the difference between the actual grayscale value and the standard grayscale value in the color space. The grayscale restoration accuracy is a percentage indicator that quantifies the camera's grayscale restoration ability, which is achieved by counting the ratio of the number of grayscale blocks that meet the standard to the total number of tested grayscale blocks.

[0062] A standard color chart is placed in a static background area, ensuring it is clearly visible in the image. Grayscale values ​​are compared block by block, and the error is determined to be below a threshold. The compliance rate is then calculated. For example, a standard grayscale chart with 16 grayscale levels can be used, combined with sub-pixel positioning technology to improve sampling accuracy. The color difference is calculated using the ΔEab formula, and accurate restoration is determined when the error is below a preset threshold. This operation, through the use of a standardized color chart and a quantified error threshold, converts grayscale restoration capability into a repeatable percentage metric, addressing the limitations of traditional methods that rely on subjective visual judgment.

[0063] The specific steps to obtain the resolution compliance rate are as follows: Identify the different line pair areas on the resolution test chart; An edge detection algorithm is used to determine the line pair clarity of each line pair area. When the grayscale jump value of the line pair edge exceeds the preset value, it is determined that the resolution of the line pair area meets the standard. The resolution compliance rate is obtained by calculating the ratio of the number of line-pair areas that meet the standards to the number of total line-pair areas. The line pair area is a black and white stripe area with a specific spatial frequency distribution on the test chart, such as the stepped line pair distribution from 20 lp / mm to 100 lp / mm on the ISO 12233 standard test chart. Line pair sharpness is a quantitative indicator that characterizes the degree of resolution between adjacent stripe boundaries and is measured by the magnitude of the grayscale jump value. The grayscale jump value is the difference in grayscale values ​​between adjacent pixels extracted by the edge detection algorithm, such as the grayscale gradient of an edge contour calculated using the Canny algorithm. The resolution compliance rate is a percentage metric that quantifies the camera's ultimate resolution capability and is calculated by the ratio of the number of compliance areas to the total number of test areas.

[0064] The line pair area on the test chart is identified and its spatial frequency parameters are extracted. The grayscale transition amplitude is calculated using an edge detection algorithm. When the transition value exceeds a preset threshold (e.g., ≥200 grayscale units), the image is considered legible. For example, sub-pixel edge location technology can be combined to improve detection accuracy, and a stepped line pair test can be used to distinguish resolution capabilities at different spatial frequencies. This operation, through binned quantitative testing and threshold determination, enables intuitive, graded assessment of resolution capabilities, offering greater engineering practicality than traditional MTF curve analysis.

[0065] The specific steps to obtain the geometric distortion rate are as follows: Identify the intersection coordinates of the distortion test grid and compare them with the theoretical intersection coordinates of the standard grid; Calculate the deviation between the actual coordinates and the theoretical coordinates of each intersection point, and obtain the geometric distortion rate by calculating the ratio of the root mean square of all deviation values ​​to the length of the grid diagonal; The distortion test grid is a standardized geometric pattern in radial or checkerboard form, whose theoretical coordinates can be predetermined using a CAD model or calibration experiment. The actual coordinates are the sub-pixel positioning results of the grid intersections identified by the image processing algorithm. The deviation value is the horizontal or vertical difference between the actual coordinates and the theoretical coordinates, and the root mean square (RMS) is the square root of the average of the squares of all deviation values. The geometric distortion rate is a dimensionless quantity that characterizes the degree of optical distortion in a lens. It is calculated by converting the ratio of the RMS deviation to the grid diagonal length into a percentage.

[0066] A sub-pixel corner detection algorithm accurately identifies grid intersection coordinates, calculates point-by-point deviations from theoretical coordinates, and uses root mean square (RMS) and normalized ratios to eliminate the effects of grid size. For example, a grid with a diagonal length of 1000 pixels can be tested. If the RMS deviation is 0.5 pixels, the distortion rate is 0.05%. This operation, through high-precision coordinate comparison and normalized calculations, quantifies optical distortion into a percentage metric consistent with visual perception, resolving the ambiguity inherent in traditional visual assessment.

[0067] The specific steps for obtaining the color restoration error value are as follows: Get the standard color space coordinates of each color block on the standard color card; Extract the actual color space coordinates of the corresponding color block in the image to be detected; The Euclidean distance between the actual coordinates and the standard coordinates of each color block is calculated, and the color restoration error value is obtained by calculating the average value of the Euclidean distances of all color blocks.

[0068] Standard color space coordinates are nominal values ​​in the CIELAB, sRGB, or AdobeRGB color spaces, typically provided by standard color chart manufacturers. Actual color space coordinates are the pixel values ​​of the corresponding color blocks extracted after image processing, requiring white balance correction and color space conversion. Euclidean distance is the geometric distance between two points in color space, such as the ΔEab value in CIELAB space. Color reproduction error is the average error metric that characterizes the camera's color reproduction accuracy. It is calculated by taking the arithmetic mean of the errors across all color blocks.

[0069] Image segmentation technology is used to locate color block areas, combined with a white balance correction algorithm to eliminate ambient light effects, and color space conversion is performed to match standard coordinates. For example, a color chart containing 12 standard color blocks can be tested, and the error value is calculated by calculating the ΔEab value for each block and taking the average. This operation, through standardized color charts and quantitative analysis, converts color reproduction capabilities into comparable numerical indicators, improving the objectivity of the evaluation compared to traditional color chart comparison methods.

[0070] The present invention quantitatively evaluates the grayscale restoration accuracy, resolution compliance rate, geometric distortion rate, and color restoration error value through standardized color card comparison, stepped line pair test, sub-pixel coordinate deviation calculation, and color space distance statistics, thereby achieving the technical effect of building a multi-dimensional static imaging performance evaluation system. Through the synergistic effect of four parameters, this system overcomes the limitations of traditional methods that rely on a single indicator or subjective judgment, and upgrades imaging quality assessment from qualitative judgment to precise grading. For example, the grayscale error can be quantified to the 1% level, the resolution difference can reach a distinction of more than 20%, the distortion rate is highly consistent with visual perception, and the color error value is benchmarked against professional display standards. Ultimately, the credibility of the comprehensive quality assessment is improved by approximately 35%, which has significant application value in scenarios such as medical imaging and remote monitoring that require high detail and color accuracy.

[0071] In one embodiment of the present invention, the specific steps of outputting the comprehensive quality score by the imaging quality assessment model are as follows: Standardize the motion characteristic parameters and static characteristic parameters and map the parameter values ​​into a preset range; Normalization is a mathematical transformation process that eliminates dimensional differences between parameters. For example, a linear mapping can be used to convert the original values ​​to the [0, 1] range. In one specific embodiment, linear normalization is used for percentage-based parameters, while piecewise linear mapping is used for absolute value parameters such as geometric distortion to adapt to their distribution characteristics.

[0072] Set the weight coefficients of motion feature parameters and static feature parameters; The weight coefficient is a numerical ratio that reflects the importance of the parameter. It is set based on the differences in the requirements of the target application scenarios. For example, the dynamic tracking performance has a higher weight in the security monitoring scenario, while the color restoration has a higher weight in the remote monitoring scenario. The weight coefficient is determined by expert experience evaluation or machine learning optimization methods, such as iterative optimization of weight combinations through genetic algorithms, or regression analysis based on user subjective scores and parameter values. The weight coefficient must meet constraints to ensure rationality, such as the sum of all weights being equal to 1 to avoid imbalance in weight distribution. In a specific embodiment, the motion blur weight is set to 0.25, the grayscale restoration accuracy weight is set to 0.15, and the trajectory deviation rate weight in dynamic scenes can be increased from 0.1 to 0.3 to enhance sensitivity.

[0073] Substitute the motion feature parameters, static feature parameters and corresponding weight coefficients into the quality calculation formula to output the comprehensive quality score.

[0074] The mass calculation formula is as follows:

[0075] in, For the overall quality rating, is the motion trajectory deviation rate, is the motion blur, is the dynamic resolution attenuation coefficient, is the grayscale restoration accuracy, is the resolution compliance rate, is the geometric distortion rate, is the color restoration error value, 、 and are the weight coefficients of motion trajectory deviation rate, motion blur, and dynamic resolution attenuation coefficient, respectively, and ; 、 、 are the weight coefficients of grayscale restoration accuracy, resolution compliance rate, geometric distortion rate, and color restoration error value, respectively, and .

[0076] This embodiment provides a method for detecting the imaging quality of home cameras. It eliminates dimensional differences through parameter standardization to ensure the rationality of weight distribution, realizes the scene adaptability of the evaluation model through flexible configuration of weight coefficients, and integrates multi-dimensional parameters through a mathematical model to generate an intuitive score. This can achieve the technical effect of improving the differentiation of different camera performances and enhancing the correlation with users' subjective evaluations, while providing an accurate parameter improvement basis for product optimization.

[0077] In one embodiment of the present invention, the specific steps for obtaining the imaging quality level of the camera are as follows: setting four quality level thresholds, which are sorted from small to large as the first threshold, the second threshold, the third threshold and the fourth threshold; when the comprehensive quality score is less than the first threshold, it is judged as an unqualified level; when the comprehensive quality score is greater than or equal to the first threshold and less than the second threshold, it is judged as a basically qualified level; when the comprehensive quality score is greater than or equal to the second threshold and less than the third threshold, it is judged as a qualified level; when the comprehensive quality score is greater than or equal to the third threshold and less than the fourth threshold, it is judged as a good level; when the comprehensive quality score is greater than or equal to the fourth threshold, it is judged as an excellent level.

[0078] The quality level threshold is a preset value used to divide the imaging quality score interval. It can be defined by industry standards or user needs. For example, the quality level threshold includes values ​​such as 60 points, 75 points, 90 points, and 95 points. This threshold sequence discretizes the continuous score interval into five quality levels through hierarchical sorting. Each level corresponds to a specific performance level. For example, "high quality" represents industry-leading performance, and "unqualified" indicates failure to meet basic requirements.

[0079] The comprehensive quality score is a quantitative evaluation result of the camera on multiple imaging parameters, which can be obtained by analyzing and calculating the original image data through an algorithm model. For example, the comprehensive quality score is generated based on the weighted average method.

[0080] The grading label determined as unqualified, basically qualified, qualified, good or excellent can be a structured identifier used to identify the performance positioning of the camera. For example, the identifier can be associated with a product specification or a user purchase guide.

[0081] The overall quality score is compared against preset thresholds in a step-by-step manner, for example, first determining whether it falls below the first threshold, and then determining whether it falls within the subsequent threshold ranges, to match the corresponding grade label. This multi-level comparison process achieves precise grading. For example, a score of 80 will trigger comparisons with the second threshold (75 points) and the third threshold (90 points), ultimately determining a passing grade. The grading results are presented in a structured report, noting the camera's compliance with standards in different dimensions, such as whether dynamic tracking performance meets standards or whether color error exceeds the threshold range.

[0082] This invention discretizes continuous scores into five quality levels by setting a stepped threshold sequence. This system, combined with a step-by-step comparison and judgment mechanism, generates structured grading results, transforming abstract comprehensive scores into intuitive grade identifiers. This solution improves assessment accuracy by quantifying cutoff points, allowing users to quickly identify product performance based on grade labels. Configurable thresholds enhance scenario adaptability, ultimately creating a multi-level quality profile that is highly correlated with actual user satisfaction, providing a clear basis for product optimization and purchasing decisions.

[0083] The present invention also provides a home camera imaging quality detection system, comprising a processor and a memory for storing instructions executable by the processor; Among them, the processor is configured to execute the instructions to implement the steps of the above-mentioned method for detecting imaging quality of a home camera.

[0084] The above description is only a preferred embodiment of the present invention. It should be pointed out that ordinary technicians in this technical field can make several improvements and modifications without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for detecting the imaging quality of a home camera, characterized in that: The following steps are involved: A preset scene containing a dynamic target area and a static background area is constructed based on a home scene, and a sequence of images to be detected is obtained by the camera in the preset scene. The preset scene includes generating motion trajectories using a dynamic target simulator and arranging static background elements to form visual features. A light gradient change control device is also introduced to simulate typical home lighting conditions. Performing region-by-region preprocessing on the image sequence to be detected, and outputting the image sequence to be detected containing a dynamic target region and a static background region after preprocessing; Extracting motion feature parameters of the dynamic target area in the image sequence to be detected after preprocessing, and extracting static feature parameters of the static background area in the image sequence to be detected after preprocessing; Input the motion feature parameters and static feature parameters into a preset imaging quality assessment model and output a comprehensive quality score; The camera's imaging quality level is obtained based on the comparison result of the comprehensive quality score and the preset threshold.

2. A method for detecting imaging quality of a home camera according to claim 1, characterized in that: The dynamic target area is provided with a plurality of standard moving objects with different motion trajectories, and the static background area is provided with a standard color card with multi-level grayscale gradients, a resolution test chart, and a distortion test grid. The specific steps for obtaining the image sequence to be detected are as follows: Control the ambient light intensity of the preset scene to change step by step according to the set gradient within the preset range, and each light intensity level maintains stable lighting for the set time; Control the standard moving object in the dynamic scene to move along a straight line, a sine curve, and a circular trajectory at different speeds, and continuously capture multiple frames of images for each motion state; The static background area remains fixed, and images are captured synchronously at each ambient light intensity level to form a sequence of images to be detected that includes different lighting conditions and motion states.

3. A method for detecting imaging quality of a home camera according to claim 1, characterized in that: The regional preprocessing specifically includes: A convolutional neural network model based on semantic segmentation is used to divide each frame of the image sequence to be detected into regions, and the pixel coordinate ranges of the dynamic target area and the static background area are determined; A denoising algorithm based on joint filtering in the spatiotemporal domain is used to eliminate motion artifacts in the dynamic target area, and an algorithm based on adaptive brightness correction is used to eliminate uneven ambient light interference in the static background area.

4. A method for detecting imaging quality of a home camera according to claim 3, characterized in that: The specific steps of using the denoising algorithm based on spatiotemporal joint filtering to eliminate motion artifacts in the dynamic target area are as follows: Calculate the grayscale change rate of each pixel point on the time axis within the dynamic target area. When the grayscale change rate exceeds the preset threshold, it is determined to be a motion artifact candidate point; A spatiotemporal joint filtering algorithm is used for the motion artifact candidate points. The spatiotemporal joint filtering algorithm uses adaptive median filtering in the spatial domain and grayscale mean weighted filtering in the temporal domain.

5. A method for detecting imaging quality of a home camera according to claim 3, characterized in that: The specific steps of using the algorithm based on adaptive brightness correction to eliminate the uneven interference of ambient light in the static background area are as follows: Extract the brightness histogram of the static background area and determine the peak interval of the brightness distribution; Based on the peak interval of the brightness distribution, a polynomial fitting algorithm is used to perform curve fitting on the brightness distribution of the static background area to obtain the brightness attenuation function; The brightness value of each pixel is corrected according to the brightness attenuation function so that the brightness standard deviation of the corrected static background area is controlled within a preset range.

6. A method for detecting imaging quality of a home camera according to claim 2, characterized in that: The motion characteristic parameters include the trajectory deviation rate, motion blur, and dynamic resolution attenuation coefficient of the standard moving body. The specific steps for obtaining the trajectory deviation rate are as follows: Generate a theoretical trajectory coordinate sequence based on the designed trajectory of the standard moving body in the preset scene; Obtain the center coordinate sequence of the standard moving object in the actual image through the target detection algorithm; Based on the theoretical trajectory coordinate sequence, the Euclidean distance between the theoretical trajectory coordinate and the actual center coordinate in each frame image is calculated; The trajectory deviation rate is obtained by calculating the ratio of the average Euclidean distance of all frames to the total length of the theoretical trajectory; The specific steps to obtain motion blur are as follows: Perform Fourier transform on the image of the dynamic target area to obtain the amplitude spectrum in the frequency domain; Calculate the energy proportion of high-frequency components in the amplitude spectrum; The motion blur is 1 minus the energy proportion of the high-frequency component; The specific steps to obtain the dynamic resolution attenuation coefficient are as follows: Setting a resolution test micro-mark on the surface of a standard moving object, wherein the micro-mark comprises black and white stripes of different spatial frequencies; The modulation transfer function of the micro-mark image in motion is calculated by Fourier transform; The dynamic resolution attenuation coefficient is obtained by calculating the ratio of the cutoff frequency of the modulation transfer function in the moving state to the cutoff frequency in the static state.

7. A method for detecting imaging quality of a home camera according to claim 2, characterized in that: The static feature parameters include grayscale restoration accuracy, resolution compliance rate, geometric distortion rate and color restoration error value. The specific steps for obtaining the grayscale restoration accuracy are as follows: Compare the standard grayscale value of each grayscale block on the standard color card with the actual grayscale value of the corresponding position in the image to be detected; Calculate the absolute error between the actual grayscale value and the standard grayscale value. When the absolute error is less than or equal to the preset value, it is determined that the restoration is accurate. The grayscale restoration accuracy is obtained by calculating the ratio of the number of accurately restored grayscale blocks to the total number of grayscale blocks; The specific steps to obtain the resolution compliance rate are as follows: Identify the different line pair areas on the resolution test chart; An edge detection algorithm is used to determine the line pair clarity of each line pair area. When the grayscale jump value of the line pair edge exceeds the preset value, it is determined that the resolution of the line pair area meets the standard. The resolution compliance rate is obtained by calculating the ratio of the number of line-pair areas that meet the standards to the number of total line-pair areas. The specific steps to obtain the geometric distortion rate are as follows: Identify the intersection coordinates of the distortion test grid and compare them with the theoretical intersection coordinates of the standard grid; Calculate the deviation between the actual coordinates and the theoretical coordinates of each intersection point, and obtain the geometric distortion rate by calculating the ratio of the root mean square of all deviation values ​​to the length of the grid diagonal; The specific steps for obtaining the color restoration error value are as follows: Get the standard color space coordinates of each color block on the standard color card; Extract the actual color space coordinates of the corresponding color block in the image to be detected; The Euclidean distance between the actual coordinates and the standard coordinates of each color block is calculated, and the color restoration error value is obtained by calculating the average value of the Euclidean distances of all color blocks.

8. The method for detecting imaging quality of a home camera according to claim 1, wherein: The specific steps of outputting the comprehensive quality score by the imaging quality assessment model are as follows: Standardize the motion characteristic parameters and static characteristic parameters and map the parameter values ​​into a preset range; Set the weight coefficients of motion feature parameters and static feature parameters; Substitute the motion feature parameters, static feature parameters and corresponding weight coefficients into the quality calculation formula to output the comprehensive quality score.

9. A method for detecting imaging quality of a home camera according to claim 1, characterized in that: The specific steps of obtaining the imaging quality level of the camera are as follows: Set four quality level thresholds, which are sorted from small to large as the first threshold, the second threshold, the third threshold and the fourth threshold; When the comprehensive quality score is less than the first threshold, it is judged as unqualified; When the comprehensive quality score is greater than or equal to the first threshold and less than the second threshold, it is judged as a basic qualified level; When the comprehensive quality score is greater than or equal to the second threshold and less than the third threshold, it is judged as a qualified grade; When the comprehensive quality score is greater than or equal to the third threshold and less than the fourth threshold, it is judged as a good grade; When the comprehensive quality score is greater than or equal to the fourth threshold, it is determined to be a high-quality grade.

10. A home camera imaging quality detection system, characterized in that: comprising a processor and a memory for storing instructions executable by said processor; The processor is configured to execute the instructions to implement the steps of a method for detecting imaging quality of a home camera as described in any one of claims 1 to 9.

Images