Method and system for optimizing management model of video image shot by unmanned aerial vehicle
By real-time monitoring and dynamic adjustment of the drone's flight status and shooting parameters, combined with image processing algorithms to optimize the quality of drone video images, the problem of insufficient environmental adaptability in traditional methods has been solved, achieving high-quality and stable image capture.
Patent Information
- Application Number
- CN202410865042.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-01
- Publication Date
- 2024-11-29
AI Technical Summary
Traditional UAV video image management methods lack adaptability to real-time environmental changes, resulting in inconsistent image quality and an inability to dynamically adjust flight and shooting parameters, thus affecting image quality and processing capabilities.
By monitoring the drone's flight status and environmental conditions in real time, the algorithm evaluates image quality using edge detection, contrast analysis, color histogram equalization, and gray-level co-occurrence matrix, dynamically adjusts exposure, ISO, aperture, and flight path, and optimizes image processing by combining noise recognition and sharpening techniques.
It improves the quality and stability of drone video images, enhances adaptability to different environments, reduces human intervention, and improves operational efficiency and image application value.
Smart Images

Figure CN119052629A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) video image management, and in particular to an optimization method and system for a management model of video images shot by an unmanned aerial vehicle (UAV). Background Art
[0002] Traditional UAV video image management methods mainly focus on capturing videos under fixed flight paths and camera settings, and rarely consider the impact of real-time environmental changes on image quality. These methods lack dynamic evaluation and adjustment of the relationship between flight status, environmental conditions and image quality, making it difficult to maintain consistent image quality under different flight conditions.
[0003] Traditional methods are unable to dynamically adjust flight and shooting parameters based on environmental changes or image quality feedback. At the same time, image quality will fluctuate under different lighting conditions and environments, affecting the final use effect, and have limited ability to handle image optimization in specific scenarios.
[0004] Therefore, there is an urgent need for an optimization method for the management model of drone video images to solve the problems of traditional methods' lack of adaptability to real-time environmental changes, large fluctuations in image quality, and limited processing capabilities, so as to further improve image quality. Summary of the invention
[0005] In view of the above existing problems, the present invention is proposed.
[0006] Therefore, the present invention provides an optimization method for a management model of video images shot by a drone, which can solve the problems of large image quality fluctuations and poor environmental adaptability of traditional static and non-adaptive methods.
[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for optimizing a management model of video images shot by a drone, comprising:
[0008] Based on video images and data, the flight status and environmental conditions of the UAV are monitored in real time, and a preliminary analysis of the video images is performed. Based on the preliminary analysis results, dynamic range adjustment is performed; video images are synthesized through software algorithms, and image quality is analyzed using noise recognition algorithms; UAV flight parameters are optimized based on image quality feedback, flight path planning parameters are set according to the target shooting area and expected image quality requirements, and the flight path is dynamically adjusted using image quality feedback; a secondary evaluation is performed, the camera shooting parameters are automatically adjusted, and the image is processed based on the content recognized by the model.
[0009] As a preferred solution of the optimization method of the management model of the drone video image shooting described in the present invention, the video images and data include: using a camera carried by a drone to capture video images on a predetermined flight path, in the process of capturing video images, real-time monitoring of the flight status and ambient lighting conditions of the drone, continuously collecting environmental data through sensors on the drone, and automatically adjusting the camera's shooting settings based on the collected environmental data, adjusting the exposure time, ISO value and aperture size.
[0010] As a preferred solution of the optimization method of the management model of the video images shot by the drone of the present invention, wherein: the preliminary analysis of the video images includes image quality assessment by edge detection algorithm and calculation of contrast index;
[0011] Use the Sobel algorithm to calculate the gradient G of image I in the x and y directions x and G y :
[0012]
[0013] According to the gradients in the x and y directions, the edge strength G of each pixel is calculated:
[0014]
[0015] The edge density D of the image is quantified by counting the proportion of pixels with edge strength higher than the threshold T in the total number of pixels:
[0016]
[0017] Introduce the Canny edge detection algorithm and apply the Gaussian filter to reduce image noise:
[0018] I′=G σ *I
[0019] Among them, I′ is the image processed by Gaussian filtering, G σ is a Gaussian kernel with standard deviation σ, * indicates the convolution operation;
[0020] Non-maximum suppression (NMS) is used to determine the true edge. For each pixel in the image, only the pixel with the local maximum value in the gradient direction is considered to be the edge. At the same time, by setting a low threshold T low and high threshold T high , the Canny algorithm distinguishes strong edges, weak edges, and non-edges; among them, a strong edge refers to a gradient with a magnitude higher than T high The weak edge refers to the gradient amplitude of T low and T highPixels between; edge tracking suppresses isolated weak edges, and only weak edges are retained when they are connected to strong edges;
[0021] The calculation of contrast index includes calculating global contrast and local contrast;
[0022] The global contrast is calculated by calculating the maximum brightness L of the image. max and minimum brightness L min quantify the difference between:
[0023] C global =L max -L min
[0024] The local contrast is to select a small window W in the image, and the local contrast C local Calculated by the standard deviation of pixel brightness within the current window:
[0025]
[0026] Where N is the number of pixels in the window, L i is the brightness of the ith pixel, is the average brightness of the pixels in the window;
[0027] Combining global and local contrast indices, an overall contrast index C is defined total To assess the contrast level of an image:
[0028]
[0029] in, is the average value of the local contrast of all windows in the image, α and β are weight coefficients used to adjust the contribution ratio of global contrast and local contrast in the overall contrast index;
[0030] High dynamic range images adjust the global contrast through dynamic range compression, applying a logarithmic transformation to reduce the contrast in highlight areas while increasing detail in dark areas:
[0031] L′=log(1+L)
[0032] Where L′ represents the adjusted brightness value, and L is the original brightness value;
[0033] The local contrast is enhanced by adaptive histogram equalization (AHE), which performs local histogram equalization on different areas of the image to improve local details:
[0034] L′ i =AHE(L i )
[0035] Among them, L′ i is the brightness value after local contrast enhancement, and AHE(·) represents the adaptive histogram equalization operation.
[0036] As a preferred solution of the optimization method of the management model of the video images shot by the drone of the present invention, wherein: the preliminary analysis of the video images also includes:
[0037] In order to evaluate the color distribution and saturation of the image, color histogram equalization and color diversity evaluation are introduced. The equalization operation is performed on the color histogram to equalize the histogram of each color channel and improve the color balance and saturation of the image:
[0038] H′(c)=CDF H (c)×(L max -L min )+L min
[0039] Among them, H′(c) is the color histogram after equalization, CDF H (c) is the cumulative distribution function of the original color histogram H, where c represents the color channel;
[0040] Assess color diversity by calculating the standard deviation of the colors in the image:
[0041]
[0042] Among them, c i is the color value of the i-th pixel in the image, is the average color value of all pixels, and M is the total number of pixels;
[0043] The gray level co-occurrence matrix GLCM is used to analyze the statistical characteristics of image texture. The texture information of the image can provide important clues to scene complexity and image quality:
[0044]
[0045] Among them, GLCM(i,j) is the value of the element in the gray-level co-occurrence matrix, which represents the frequency of pixels adjacent to pixel value i having value j;
[0046] The consistency of image texture is evaluated by calculating the uniformity of the GLCM:
[0047]
[0048] By analyzing the characteristics of GLCM in different directions, the directionality of image texture is evaluated;
[0049] Output preliminary analysis results considering edge detection, contrast analysis, and color distribution and texture characteristics:
[0050] Q total =ω1·D+ω2·C total +ω3·σ color +ω4·Contrast GLCM +ω5
[0051] Homogeneity GLCM
[0052] Among them, Q total is the comprehensive score of image quality, and ω1, ω2, ω3, ω4 and ω5 are weight coefficients for adjusting the contribution of each score.
[0053] As a preferred solution of the optimization method of the management model of the video images taken by the drone of the present invention, wherein: the execution of dynamic range adjustment includes setting the scoring level threshold, wherein the low scoring threshold T' low Used to distinguish low-quality images from medium-quality images. total Less than T' low , the image quality is considered low:
[0054] T′ low =δ·Median(Q total )-ε
[0055] Among them, δ and ε are adjustment factors, Median(Q total ) is the image Q total The median of the values;
[0056] High score threshold T' high Used to distinguish between medium-quality images and high-quality images. total Greater than T' high , the image quality is considered high:
[0057] T′ high =δ·Median(Q total )+θ
[0058] where θ is another adjustment factor that defines how much Q exceeds the median total Values are considered high quality;
[0059] The middle score interval is defined as Q total The value is at T' low and T' high between:
[0060] T′ low total <T′ high
[0061] Images in the medium rating range were considered to be of medium quality;
[0062] If the score result is in the low score range and the image is underexposed, the system automatically adjusts the settings, increases the ISO value and reduces the shutter speed or selects a larger aperture value, and the f-number decreases. If the score result is in the low score range and the image is overexposed, it means that the pixel values are concentrated in the high-brightness area, the system automatically reduces the ISO value, increases the shutter speed or selects a smaller aperture value, and the f-number increases.
[0063] If a deviation is found by comparing the average brightness value of the calculated image with the preset ideal brightness range, it means that the exposure setting needs to be adjusted; when special objects need to be highlighted in the image, the system uses program-controlled aperture to achieve changes in depth of field and create a background blur effect, regardless of the specific content or quality of the background. Furthermore, the system uses edge detection algorithms to enhance the edge contours of special objects and draws a dotted frame around them for identification. The marking operation is independent of the nature of the object or other elements of the image;
[0064] When a special object needs to be visually magnified to emphasize its location in the image, the system digitally magnifies the selected area according to a pre-set magnification ratio, based only on the object's initial size and location information; if the magnification operation causes a change in pixel resolution, the system will execute a sharpening algorithm to adjust the clarity of the magnified area, regardless of the overall quality of the image or other details;
[0065] If real-time analysis indicates that the image histogram tends to the low-brightness area, indicating that the lighting conditions may lead to underexposure, the system will automatically increase the ISO value and reduce the shutter speed, or adjust the aperture to a larger value to absorb more light; when the histogram tends to the high-brightness area, suggesting that there is a risk of overexposure, the system will reduce the ISO value, increase the shutter speed or select a smaller aperture value to limit the entry of light.
[0066] If a special object needs to be highlighted in the image, but the background information needs to be kept to provide context, the system will slightly adjust the aperture to a medium size to achieve a slight background blur effect. Once the background blur is complete, if the edge of the special object needs to be enhanced to improve its definition in the image, an edge enhancement filter is applied and a dotted box is drawn around the object to mark it. The marking process takes into account the overall style of the image to ensure that it matches the background blur strategy.
[0067] When a special object needs to be more visually obvious due to its importance, the system digitally enlarges the target object according to the predetermined enlargement ratio, and adjusts the sharpness of the enlarged area to adapt to the resolution change after the enlargement; if the enlargement result causes the details of the special object to be unclear, the details will be sharpened, but over-sharpening will be avoided to avoid introducing artificial noise. The sharpening operation is limited to maintaining the clarity of the enlarged details without changing other parts of the image;
[0068] If the score is in the high score range, it means that the image analysis shows a slight color temperature deviation, and the system automatically adjusts the white balance parameters based on an accurate assessment of the current image color distribution to match the natural light color temperature; when the image shows a slight exposure deviation, the system fine-tunes the exposure compensation value to cope with subtle lighting changes, keeping the adjustment range to a minimum to avoid significantly changing the overall image exposure effect;
[0069] If a special object needs to be slightly highlighted in the image, but the background information needs to be kept intact, the aperture size is slightly adjusted to minimize the background blur; when a special object needs to be carefully marked to improve recognition, a non-invasive marking method is used to slightly adjust the brightness or contrast or draw a subtle border to avoid the marking process interfering with the main visual content of the image;
[0070] When the details of a special object are important to the overall understanding but it is necessary to avoid disrupting the image balance, slightly enlarge the target object, carefully control the proportions during the enlargement process to ensure that it does not affect the overall composition of the image or cause unnecessary cropping; if enlargement results in loss of details of the special object, apply detailed sharpening processing and adjust the enlarged area, keeping other operations unchanged.
[0071] As a preferred solution of the optimization method of the management model of the video images shot by the drone of the present invention, wherein: the noise recognition algorithm analyzes the image including converting the image into the frequency domain to identify the noise:
[0072]
[0073] Among them, f(x,y) is the representation of the original image in the spatial domain, and F(u,v) is the representation of the image in the frequency domain. represents Fourier transform;
[0074] Perform waveform threshold processing. In the frequency domain, identify and process noise according to the set threshold T:
[0075]
[0076] Where F′(u,v) is the processed frequency domain representation, in which the frequency components below the threshold T are considered as noise and set to 0;
[0077] By performing an inverse Fourier transform, the noise-processed frequency domain image is converted back to the spatial domain:
[0078]
[0079] in, represents the inverse Fourier transform, and f′(x,y) is the representation of the denoised image in the spatial domain;
[0080] Apply the Unsharp Mask technique to enhance the edges and details of the image:
[0081] g(x,y)=f′(x,y)+k·(f′(x,y)-B(x,y))
[0082] Among them, g(x,y) is the sharpened image, B(x,y) is the blurred image obtained by Gaussian blurring the original image f′(x,y), and k is the enhancement coefficient used to adjust the intensity of sharpening;
[0083] An adaptive threshold setting is introduced for waveform threshold processing, and the threshold is dynamically adjusted based on the local frequency domain characteristics of the image:
[0084] T adaptive (u,v)=θ·std(F local (u,v))+ρ
[0085] Among them, T adaptive (u,v) is the adaptive threshold at the frequency domain coordinate (u,v), F local (u, v) represents the frequency components in the local frequency domain, std is the standard deviation function used to evaluate the variability in the local frequency domain, and θ and ρ are adjustment coefficients;
[0086] Perform local noise suppression in the frequency domain rather than globally to preserve finer details:
[0087]
[0088] Noise components are suppressed in the frequency domain based on the ratio of their intensity relative to a local adaptive threshold, allowing for more flexible processing of noise in different frequency domain regions;
[0089] Use edge detection algorithms to guide the sharpening process so that the sharpening is mainly concentrated in the edge area and the impact on the smooth area is reduced:
[0090] g′(x,y)=g(x,y)+λ·(g(x,y)-g smooth (x,y))·E(x,y)
[0091] Among them, g smooth(x, y) is the result of Gaussian blurring of image g(x, y), E(x, y) is the edge intensity map obtained from the original image f′(x, y), and λ is an adjustment factor that controls the adaptive adjustment of sharpening intensity as the edge intensity changes;
[0092] Through edge-aware sharpening technology, the sharpening effect is enhanced in the edge area of the image and naturally suppressed in the non-edge smooth area, thereby minimizing the artifacts and noise enhancement that may be generated during the sharpening process.
[0093] As a preferred solution of the optimization method of the management model of the video images shot by the drone of the present invention, wherein: the dynamic adjustment of the flight path by using image quality feedback includes, for the video images that have been evaluated as low scores, the system determines that the problem is underexposure or overexposure through real-time analysis of the captured images, and adjusts the flight parameters; when the video image is underexposed, the system increases the ISO value, reduces the aperture f-number, and reduces the shutter speed to introduce a large amount of light; if the image quality is not significantly improved, the system will re-plan the flight path, adjust the height of the drone or change the angle to obtain the best lighting conditions or avoid backlight; the system will continuously monitor the image quality and continue to adjust the flight speed and camera parameters as needed until a satisfactory image quality level is achieved;
[0094] For video images that were once assessed as medium, the system fine-tunes the camera's ISO, aperture, and shutter speed. Considering the importance of shooting angle and timing to image quality, the system fine-tunes the drone's flight path based on real-time feedback to find the optimal shooting position and angle, while also considering the shooting time to take advantage of the best natural lighting conditions, while emphasizing the use of the golden hour's lighting advantages and avoiding backlight or excessive direct light;
[0095] For video images that have been evaluated as highly rated, the system first ensures flight stability by monitoring flight stability and fine-tuning flight parameters to adapt to subtle changes in ambient lighting. This prevents unnecessary blur or vibration introduced by flight operations. For images that already have high quality, the system accurately monitors slight changes in lighting conditions and adjusts white balance and exposure compensation in a timely manner to maintain color accuracy and the overall brightness balance of the image. It also slightly adjusts flight altitude and speed as needed and optimizes shooting angles to ensure that the best viewing angle and lighting effects are captured.
[0096] Another object of the present invention is to provide a robotic arm control system based on a visual system, which solves the problems of unstable image quality and insufficient adaptability to environmental changes through real-time monitoring and feedback mechanism, dynamic adjustment of flight and shooting parameters, and image processing technology.
[0097] As a preferred solution of the optimization system of the management model of the drone video image shooting described in the present invention, the system includes an image acquisition module, an image processing module, a drone control module, a user interface module, a path planning module and a dynamic adjustment module; the image acquisition module is equipped with a high dynamic range HDR sensor for acquiring video frames with different exposure levels; the image processing module is equipped with an HDR image synthesis algorithm, a real-time noise suppression algorithm, an automatic exposure adjustment algorithm and a color correction algorithm; wherein the HDR image synthesis algorithm processes the video frames with different exposure levels acquired by the image acquisition module to generate HDR video frames; the real-time noise suppression algorithm is used to reduce the noise in the video frames acquired under low light conditions; the automatic exposure adjustment algorithm is used to automatically adjust the exposure parameters according to real-time video analysis number; the color correction algorithm is used to optimize the color performance of the video frame; the drone control module is configured with an adaptive flight strategy logic unit and an environmental perception unit, which can adjust the drone's flight parameters according to the analysis results of the image processing module, and use the light sensor data to assist in adjusting the flight and shooting parameters; the user interface module provides an operation interface that allows the user to manually adjust the image acquisition parameters, view the real-time video preview, and receive the parameter adjustment recommended by the system; the path planning module is used to optimize the drone's flight parameters according to image quality feedback, set the flight path planning parameters according to the target shooting area and the expected image quality requirements, and dynamically adjust the flight path using image quality feedback; the dynamic adjustment module is used to perform secondary evaluation, automatically adjust the camera shooting parameters, and process the image based on the content recognized by the model.
[0098] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the optimization method of the management model of video images shot by a drone as described above are implemented.
[0099] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for optimizing a management model for video images shot by a drone as described above.
[0100] Beneficial effects of the present invention: The method of the present invention improves the management and processing capabilities of drone video images. By real-time monitoring of the drone's flight status and environmental conditions and combining image processing technology, dynamic adjustment of the flight path and camera settings is achieved. Not only does it enhance the adaptability to different environmental conditions and ensure that high-quality images can be obtained under various lighting and environments, but it also optimizes operational efficiency and reduces reliance on manual intervention. Through automated image quality assessment and adjustment mechanisms, the applicability and effectiveness of drones in various application scenarios such as surveillance, surveying, and media production are improved, making drone video shooting more flexible and efficient, and able to produce clearer and more stable images, greatly improving the application value of images and user experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0101] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.
[0102] Figure 1 A schematic flow chart of a method for optimizing a management model for video images shot by a drone provided in accordance with an embodiment of the present invention.
[0103] Figure 2 A schematic diagram of a module of an optimization system for a management model of video images shot by a drone provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0104] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.
[0105] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0106] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0107] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.
[0108] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.
[0109] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0110] Example 1
[0111] Reference Figure 1 , which is the first embodiment of the present invention, provides a method for optimizing a management model of video images shot by a drone, comprising:
[0112] S1: Based on the video images and data, the flight status and environmental conditions of the drone are monitored in real time, and the video images are preliminarily analyzed. Based on the preliminary analysis results, dynamic range adjustment is performed.
[0113] The video images and data include capturing video images on a predetermined flight path using a camera mounted on a drone, real-time monitoring of the drone's flight status and ambient lighting conditions during the process of capturing the video images, continuously collecting environmental data through sensors on the drone, and automatically adjusting the camera's shooting settings based on the collected environmental data, adjusting exposure time, ISO value, and aperture size.
[0114] The preliminary analysis of the video image includes evaluating the image quality by edge detection algorithm and calculating contrast index;
[0115] Use the Sobel algorithm to calculate the gradient G of image I in the x and y directions x and G y :
[0116]
[0117] According to the gradients in the x and y directions, the edge strength G of each pixel is calculated:
[0118]
[0119] The edge density D of the image is quantified by counting the proportion of pixels with edge strength higher than the threshold T in the total number of pixels:
[0120]
[0121] Introduce the Canny edge detection algorithm and apply the Gaussian filter to reduce image noise:
[0122] I′=G σ *I
[0123] Among them, I′ is the image processed by Gaussian filtering, G σ is a Gaussian kernel with standard deviation σ, * indicates the convolution operation;
[0124] Non-maximum suppression (NMS) is used to determine the true edge. For each pixel in the image, only the pixel with the local maximum value in the gradient direction is considered to be the edge. At the same time, by setting a low threshold T low and high threshold T high , the Canny algorithm distinguishes strong edges, weak edges, and non-edges; among them, a strong edge refers to a gradient with a magnitude higher than T high The weak edge refers to the gradient amplitude of T low and T high Pixels between; edge tracking suppresses isolated weak edges, and only weak edges are retained when they are connected to strong edges;
[0125] The calculation of contrast index includes calculating global contrast and local contrast;
[0126] The global contrast is calculated by calculating the maximum brightness L of the image. max and minimum brightness L min quantify the difference between:
[0127] C global =L max -L min
[0128] The local contrast is to select a small window W in the image, and the local contrast C local Calculated by the standard deviation of pixel brightness within the current window:
[0129]
[0130] Where N is the number of pixels in the window, L i is the brightness of the ith pixel, is the average brightness of the pixels in the window;
[0131] Combining global and local contrast indices, an overall contrast index C is defined total To assess the contrast level of an image:
[0132]
[0133] in, is the average value of the local contrast of all windows in the image, α and β are weight coefficients used to adjust the contribution ratio of global contrast and local contrast in the overall contrast index; high dynamic range images adjust the global contrast by dynamic range compression, and apply logarithmic transformation to reduce the contrast of highlight areas while increasing the details of dark areas:
[0134] L′=log(1+L)
[0135] Where L′ represents the adjusted brightness value, and L is the original brightness value;
[0136] The local contrast is enhanced by adaptive histogram equalization (AHE), which performs local histogram equalization on different areas of the image to improve local details:
[0137] L′ i =AHE(L i )
[0138] Among them, L′ i is the brightness value after local contrast enhancement, and AHE(·) represents the adaptive histogram equalization operation.
[0139] In order to evaluate the color distribution and saturation of the image, color histogram equalization and color diversity evaluation are introduced. The equalization operation is performed on the color histogram to equalize the histogram of each color channel and improve the color balance and saturation of the image:
[0140] H′(c)=CDF H (c)×(L max -L min )+L min
[0141] Among them, H′(c) is the color histogram after equalization, CDF H (c) is the cumulative distribution function of the original color histogram H, where c represents the color channel;
[0142] Assess color diversity by calculating the standard deviation of the colors in the image:
[0143]
[0144] Among them, c i is the color value of the i-th pixel in the image, is the average color value of all pixels, and M is the total number of pixels;
[0145] The gray level co-occurrence matrix GLCM is used to analyze the statistical characteristics of image texture. The texture information of the image can provide important clues to scene complexity and image quality:
[0146]
[0147] Among them, GLCM(m,j) is the value of the element in the gray-level co-occurrence matrix, which represents the frequency of pixels adjacent to pixel value m having value j;
[0148] The consistency of image texture is evaluated by calculating the uniformity of the GLCM:
[0149]
[0150] By analyzing the characteristics of GLCM in different directions, the directionality of image texture is evaluated;
[0151] Output preliminary analysis results considering edge detection, contrast analysis, and color distribution and texture characteristics:
[0152] Q total =ω1·D+ω2·C total +ω3·σ color +ω4·Contrast GLCM +ω5
[0153] Homogeneity GLCM
[0154] Among them, Q total is the comprehensive score of image quality, ω1, ω 2 , ω3, ω4 and ω5 are the weight coefficients for adjusting the contribution of each score.
[0155] The dynamic range adjustment includes setting a score level threshold, wherein the low score threshold T' low Used to distinguish low-quality images from medium-quality images.total Less than T' low , the image quality is considered low:
[0156] T′ low =δ·Median(Q total )-ε
[0157] Among them, δ and ε are adjustment factors, Median(Q total ) is a series of images Q total The median of the values;
[0158] High score threshold T' high Used to distinguish between medium-quality images and high-quality images. total Greater than T' high , the image quality is considered high:
[0159] T′ high =δ·Median(Q total )+θ
[0160] where θ is another adjustment factor that defines how much Q exceeds the median total Values are considered high quality;
[0161] The middle score interval is defined as Q total The value is at T' low and T' high between:
[0162] T′ low total <T′ high
[0163] Images in the medium rating range were considered to be of medium quality;
[0164] If the score result is in the low score range and the image is underexposed, the system automatically adjusts the settings, increases the ISO value and reduces the shutter speed or selects a larger aperture value, and the f-number decreases. If the score result is in the low score range and the image is overexposed, it means that the pixel values are concentrated in the high-brightness area, the system automatically reduces the ISO value, increases the shutter speed or selects a smaller aperture value, and the f-number increases.
[0165] If a deviation is found by comparing the average brightness value of the calculated image with the preset ideal brightness range, it means that the exposure setting needs to be adjusted; when special objects need to be highlighted in the image, the system uses program-controlled aperture to achieve changes in depth of field and create a background blur effect, regardless of the specific content or quality of the background. Furthermore, the system uses edge detection algorithms to enhance the edge contours of special objects and draws a dotted frame around them for identification. The marking operation is independent of the nature of the object or other elements of the image;
[0166] When a special object needs to be visually magnified to emphasize its location in the image, the system digitally magnifies the selected area according to a pre-set magnification ratio, based only on the object's initial size and location information; if the magnification operation causes a change in pixel resolution, the system will execute a sharpening algorithm to adjust the clarity of the magnified area, regardless of the overall quality of the image or other details;
[0167] If real-time analysis indicates that the image histogram tends to the low-brightness area, indicating that the lighting conditions may lead to underexposure, the system will automatically increase the ISO value and reduce the shutter speed, or adjust the aperture to a larger value to absorb more light; when the histogram tends to the high-brightness area, suggesting that there is a risk of overexposure, the system will reduce the ISO value, increase the shutter speed or select a smaller aperture value to limit the entry of light.
[0168] If a special object needs to be highlighted in the image, but the background information needs to be kept to provide context, the system will slightly adjust the aperture to a medium size to achieve a slight background blur effect. Once the background blur is complete, if the edge of the special object needs to be enhanced to improve its definition in the image, an edge enhancement filter is applied and a dotted box is drawn around the object to mark it. The marking process takes into account the overall style of the image to ensure that it matches the background blur strategy.
[0169] When a special object needs to be more visually obvious due to its importance, the system digitally enlarges the target object according to the predetermined enlargement ratio, and adjusts the sharpness of the enlarged area to adapt to the resolution change after the enlargement; if the enlargement result causes the details of the special object to be unclear, the details will be sharpened, but over-sharpening will be avoided to avoid introducing artificial noise. The sharpening operation is limited to maintaining the clarity of the enlarged details without changing other parts of the image;
[0170] If the score is in the high score range, it means that the image analysis shows a slight color temperature deviation, and the system automatically adjusts the white balance parameters based on an accurate assessment of the current image color distribution to match the natural light color temperature; when the image shows a slight exposure deviation, the system fine-tunes the exposure compensation value to cope with subtle lighting changes, keeping the adjustment range to a minimum to avoid significantly changing the overall image exposure effect;
[0171] If a special object needs to be slightly highlighted in the image, but the background information needs to be kept intact, the aperture size is slightly adjusted to minimize the background blur; when a special object needs to be carefully marked to improve recognition, a non-invasive marking method is used to slightly adjust the brightness or contrast or draw a subtle border to avoid the marking process interfering with the main visual content of the image;
[0172] When the details of a special object are important to the overall understanding but it is necessary to avoid disrupting the image balance, slightly enlarge the target object, carefully control the proportions during the enlargement process to ensure that it does not affect the overall composition of the image or cause unnecessary cropping; if enlargement results in loss of details of the special object, apply detailed sharpening processing and adjust the enlarged area, keeping other operations unchanged.
[0173] S2: Synthesize video images through software algorithms and analyze image quality using noise recognition algorithms.
[0174] The noise identification algorithm analyzes the image including, the image is converted to the frequency domain to identify the noise:
[0175]
[0176] Among them, f(x,y) is the representation of the original image in the spatial domain, and F(u,v) is the representation of the image in the frequency domain. represents Fourier transform;
[0177] Perform waveform threshold processing. In the frequency domain, identify and process noise according to the set threshold T:
[0178]
[0179] Where F′(u,v) is the processed frequency domain representation, in which the frequency components below the threshold T are considered as noise and set to 0;
[0180] By performing an inverse Fourier transform, the noise-processed frequency domain image is converted back to the spatial domain:
[0181]
[0182] in, represents the inverse Fourier transform, and f′(x,y) is the representation of the denoised image in the spatial domain;
[0183] Apply the Unsharp Mask technique to enhance the edges and details of the image:
[0184] g(x,y)=f′(x,y)+k·(f′(x,y)-B(x,y))
[0185] Among them, g(x,y) is the sharpened image, B(x,y) is the blurred image obtained by Gaussian blurring the original image f′(x,y), and k is the enhancement coefficient used to adjust the intensity of sharpening;
[0186] An adaptive threshold setting is introduced for waveform threshold processing, and the threshold is dynamically adjusted based on the local frequency domain characteristics of the image:
[0187] T adaptive(u,v)=θ·std(F local (u,v))+ρ
[0188] Among them, T adaptive (u,v) is the adaptive threshold at the frequency domain coordinate (u,v), F local (u, v) represents the frequency components in the local frequency domain, std is the standard deviation function used to evaluate the variability in the local frequency domain, and θ and ρ are adjustment coefficients;
[0189] Perform local noise suppression in the frequency domain rather than globally to preserve finer details:
[0190]
[0191] Noise components are suppressed in the frequency domain based on the ratio of their intensity relative to a local adaptive threshold, allowing for more flexible processing of noise in different frequency domain regions;
[0192] Use edge detection algorithms to guide the sharpening process so that the sharpening is mainly concentrated in the edge area and the impact on the smooth area is reduced:
[0193] g′(x,y)=g(x,y)+λ·(g(x,y)-g smooth (x,y))·E(x,y)
[0194] Among them, g smooth (x, y) is the result of Gaussian blurring of image g(x, y), E(x, y) is the edge intensity map obtained from the original image f′(x, y), and λ is an adjustment factor that controls the adaptive adjustment of sharpening intensity as the edge intensity changes;
[0195] Through edge-aware sharpening technology, the sharpening effect is enhanced in the edge area of the image and naturally suppressed in the non-edge smooth area, thereby minimizing the artifacts and noise enhancement that may be generated during the sharpening process.
[0196] S3: Optimize UAV flight parameters based on image quality feedback, set flight path planning parameters according to the target shooting area and expected image quality requirements, and dynamically adjust the flight path using image quality feedback.
[0197] For video images that were once evaluated as low scores, the system analyzes the captured images in real time, determines that the problem is underexposure or overexposure, and adjusts the flight parameters; when the video image is underexposed, the system increases the ISO value, reduces the aperture f-number, and reduces the shutter speed to introduce a large amount of light; if the image quality does not improve significantly, the system will re-plan the flight path, adjust the height of the drone or change the angle to obtain the best lighting conditions or avoid backlight; the system will continue to monitor the image quality and continue to adjust the flight speed and camera parameters as needed until a satisfactory image quality level is achieved;
[0198] For video images that were once assessed as medium, the system fine-tunes the camera's ISO, aperture, and shutter speed. Considering the importance of shooting angle and timing to image quality, the system fine-tunes the drone's flight path based on real-time feedback to find the optimal shooting position and angle, while also considering the shooting time to take advantage of the best natural lighting conditions, while emphasizing the use of the golden hour's lighting advantages and avoiding backlight or excessive direct light;
[0199] For video images that have been evaluated as highly rated, the system first ensures flight stability by monitoring flight stability and fine-tuning flight parameters to adapt to subtle changes in ambient lighting. This prevents unnecessary blur or vibration introduced by flight operations. For images that already have high quality, the system accurately monitors slight changes in lighting conditions and adjusts white balance and exposure compensation in a timely manner to maintain color accuracy and the overall brightness balance of the image. It also slightly adjusts flight altitude and speed as needed and optimizes shooting angles to ensure that the best viewing angle and lighting effects are captured.
[0200] S4: Perform secondary evaluation, automatically adjust camera shooting parameters, and process the image based on the content recognized by the model.
[0201] Example 2
[0202] The second embodiment of the present invention provides an optimization method for a management model of video images shot by a drone. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.
[0203] The effects of the method of the present invention and the prior art on the quality of UAV video images under different environmental conditions were compared. Two UAV systems were used in the experiment: one equipped with the method of the present invention for shooting, and the other equipped with a standard image processing system for shooting as a control.
[0204] Three typical environments were selected for testing to evaluate the performance of each system under different conditions, including:
[0205] Strong light environment: simulates direct sunlight conditions to test the system's ability to handle overexposure of images under strong light.
[0206] Low-light environment: simulates nighttime or indoor low-light environments to evaluate the system's image enhancement effect under low-light conditions.
[0207] Dynamic Environments: Test the system’s dynamic image stabilization and processing capabilities by operating the drone in windy conditions or photographing moving objects.
[0208] In each environment, five flight shooting missions were performed using the method of the present invention and the traditional method, and a series of video images were captured in each mission. The captured images were analyzed using image evaluation software, and the performance differences of the two methods in image quality management and processing were evaluated and compared by analyzing parameters.
[0209] After each flight mission, the captured video images were imported into the image evaluation software to score the images under each environmental condition. The experimental data are shown in Table 1.
[0210] Table 1 Comparison of experimental results
[0211]
[0212]
[0213] In all test environments, the scores of the method of the present invention in terms of image clarity, contrast, and color saturation are significantly higher than those of the traditional methods, and the noise level is also significantly lower than that of the traditional methods. When processing video images under different environmental conditions, the method of the present invention can effectively improve the quality and visual effects of the image.
[0214] Especially in dynamic environments, the method of the present invention demonstrates its superior image stabilization and processing capabilities, and can maintain image clarity and low noise levels even in windy conditions or when shooting moving objects. This is of great value for drones in application scenarios such as surveillance, surveying, and media production, as these scenarios often require operation in complex or unstable environmental conditions.
[0215] The superiority of the method of the present invention is not only reflected in the improvement of the comprehensive quality of the image, but also in its enhanced environmental adaptability and operating efficiency. Compared with the traditional method, the method of the present invention reduces the dependence on manual intervention, and through the automated image quality evaluation and adjustment mechanism, it greatly improves the processing efficiency and effect, further proving the creativity and novelty of the present invention based on the existing technology.
[0216] In summary, this embodiment clearly demonstrates the significant advantages and innovations of the present invention over the prior art through comparative analysis of objective data. The present invention can not only significantly improve the quality of drone video images in a variety of environments, but also enhance the adaptability of the system and reduce manual intervention, providing an efficient and automated solution for the management and processing of drone video images.
[0217] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
[0218] Example 3
[0219] Reference Figure 2 , which is the first embodiment of the present invention, provides an optimization system for a management model of video images taken by a drone, including:
[0220] The system includes an image acquisition module, an image processing module, a drone control module, a user interface module, a path planning module and a dynamic adjustment module;
[0221] The image acquisition module is equipped with a high dynamic range (HDR) sensor for acquiring video frames of different exposure levels; the image processing module is equipped with an HDR image synthesis algorithm, a real-time noise suppression algorithm, an automatic exposure adjustment algorithm and a color correction algorithm; wherein the HDR image synthesis algorithm processes video frames of different exposure levels acquired by the image acquisition module to generate HDR video frames; the real-time noise suppression algorithm is used to reduce noise in video frames acquired under low light conditions; the automatic exposure adjustment algorithm is used to automatically adjust exposure parameters according to real-time video analysis; the color correction algorithm is used to optimize the color performance of video frames; the drone control module is equipped with an adaptive flight strategy logic unit and an environmental perception unit, which can adjust the flight parameters of the drone according to the analysis results of the image processing module, and use the light sensor data to assist in adjusting the flight and shooting parameters; the user interface module provides an operation interface that allows the user to manually adjust the image acquisition parameters, view the real-time video preview and receive the parameter adjustment recommended by the system; the path planning module is used to optimize the drone flight parameters according to image quality feedback, set the flight path planning parameters according to the target shooting area and the expected image quality requirements, and dynamically adjust the flight path using image quality feedback; the dynamic adjustment module is used to perform secondary evaluation, automatically adjust the camera shooting parameters, and process the image based on the content recognized by the model.
[0222] Example 4
[0223] The fourth embodiment of the present invention is different from the first three embodiments in that:
[0224] If the functions are implemented in the form of software functional units and sold or used as independent products, they 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 enabling 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 described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.
[0225] Those skilled in the art will appreciate that the embodiments of the present application can provide methods, systems or computer program products. Therefore, the present application can adopt the form of complete hardware embodiment, complete software embodiment, or the embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiment of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.
[0226] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0227] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0228] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0229] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.
[0230] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
Claims
1. An optimization method for the management model of video images taken by drones, Features: include, Based on the video images and data, the flight status and environmental conditions of the drone are monitored in real time, and the video images are preliminarily analyzed. Based on the preliminary analysis results, dynamic range adjustment is performed; Synthesize video images through software algorithms and analyze image quality using noise recognition algorithms; Optimize UAV flight parameters based on image quality feedback, set flight path planning parameters according to the target shooting area and expected image quality requirements, and dynamically adjust the flight path using image quality feedback; Perform secondary evaluation, automatically adjust camera shooting parameters, and process images based on what the model recognizes.
2. The method for optimizing the management model of video images shot by a drone as claimed in claim 1, Features: The video images and data include capturing video images on a predetermined flight path using a camera mounted on the drone, real-time monitoring of the drone's flight status and ambient lighting conditions during the process of capturing the video images, continuous collection of environmental data through sensors on the drone, and automatic adjustment of the camera's shooting settings based on the collected environmental data, including exposure time, ISO value, and aperture size.
3. The method for optimizing the management model of video images shot by a drone as claimed in claim 2, Features: The preliminary analysis of the video image includes evaluating the image quality by edge detection algorithm and calculating contrast index; Use the Sobel algorithm to calculate the gradient G of image I in the x and y directions x and G y : According to the gradients in the x and y directions, the edge strength G of each pixel is calculated: The edge density D of the image is quantified by counting the proportion of pixels with edge strength higher than the threshold T in the total number of pixels: Introduce the Canny edge detection algorithm and apply the Gaussian filter to reduce image noise: I′=G σ *I Among them, I′ is the image processed by Gaussian filtering, G σ is a Gaussian kernel with standard deviation σ, * indicates the convolution operation; Non-maximum suppression (NMS) is used to determine the true edge. For each pixel in the image, only the pixel with the local maximum value in the gradient direction is considered to be the edge. At the same time, by setting a low threshold T low and high threshold T high , the Canny algorithm distinguishes strong edges, weak edges, and non-edges; among them, a strong edge refers to a gradient with a magnitude higher than T high The weak edge refers to the gradient amplitude of T low and T high Pixels between; edge tracking suppresses isolated weak edges, and only weak edges are retained when they are connected to strong edges; The calculation of contrast index includes calculating global contrast and local contrast; The global contrast is calculated by calculating the maximum brightness L of the image. max and minimum brightness L min quantify the difference between: C global =L max -L min The local contrast is to select a small window W in the image, and the local contrast C local Calculated by the standard deviation of pixel brightness within the current window: Where N is the number of pixels in the window, L i is the brightness of the ith pixel, is the average brightness of the pixels in the window; Combining global and local contrast indices, an overall contrast index C is defined total To assess the contrast level of an image: in, is the average value of the local contrast of all windows in the image, α and β are weight coefficients used to adjust the contribution ratio of global contrast and local contrast in the overall contrast index; High dynamic range images adjust the global contrast through dynamic range compression, applying a logarithmic transformation to reduce the contrast in highlight areas while increasing detail in dark areas: L′=log(1+L) Where L′ represents the adjusted brightness value, and L is the original brightness value; The local contrast is enhanced by adaptive histogram equalization (AHE), which performs local histogram equalization on different areas of the image to improve local details: L' i =AHE(L i ) Among them, L′ i is the brightness value after local contrast enhancement, and AHE(·) represents the adaptive histogram equalization operation.
4. The method for optimizing the management model of video images shot by a drone as claimed in claim 3, Features: The preliminary analysis of the video image also includes: In order to evaluate the color distribution and saturation of the image, color histogram equalization and color diversity evaluation are introduced. The color histogram is equalized to equalize the histogram of each color channel and improve the color balance and saturation of the image: H′(c)=CDF H (c)×(L max -L min )+L min Among them, H′(c) is the color histogram after equalization, CDF H (c) is the cumulative distribution function of the original color histogram H, where c represents the color channel; Assess color diversity by calculating the standard deviation of the colors in the image: Among them, c i is the color value of the i-th pixel in the image, is the average color value of all pixels, and M is the total number of pixels; The gray level co-occurrence matrix GLCM is used to analyze the statistical characteristics of image texture. The texture information of the image can provide important clues to scene complexity and image quality: Among them, GLCM(m,j) is the value of the element in the gray-level co-occurrence matrix, which represents the frequency of pixels adjacent to pixel value m having value j; The consistency of image texture is evaluated by calculating the uniformity of the GLCM: By analyzing the characteristics of GLCM in different directions, the directionality of image texture is evaluated; Output preliminary analysis results considering edge detection, contrast analysis, and color distribution and texture characteristics: Q total =ω 1 ·D+ω 2 ·C total +oh 3 ·s color +oh 4 ·Contrast GLCM +oh 5 ·Homogeneity GLCM Among them, Q total is the comprehensive score of image quality, ω 1 ,ω 2 ,ω 3 ,ω 4 and 5 It is the weight coefficient for adjusting the contribution of each score.
5. The method for optimizing the management model of video images shot by a drone as claimed in claim 4, Features: The dynamic range adjustment includes setting a score level threshold, wherein the low score threshold T' low Used to distinguish low-quality images from medium-quality images. total Less than T' low , the image quality is considered low: T′ low =δ·Median(Q total )-e Among them, δ and ε are adjustment factors, Median(Q total ) is the image Q total The median of the values; High score threshold T' high Used to distinguish between medium-quality images and high-quality images. total Greater than T' high , the image quality is considered high: T′ high =δ·Median(Q total )+θ Among them, θ is another adjustment factor used to determine the Q range that exceeds the median total Values are considered high quality; The middle score interval is defined as Q total The value is at T' low and T' high between: T′ low <Q total <T high Images in the medium rating range were considered to be of medium quality; If the score result is in the low score range and the image is underexposed, the system automatically adjusts the settings, increases the ISO value and reduces the shutter speed or selects a larger aperture value, and the f-number decreases. If the score result is in the low score range and the image is overexposed, it means that the pixel values are concentrated in the high-brightness area, the system automatically reduces the ISO value, increases the shutter speed or selects a smaller aperture value, and the f-number increases. If a deviation is found by comparing the average brightness value of the calculated image with the preset ideal brightness range, it means that the exposure setting needs to be adjusted; when special objects need to be highlighted in the image, the system uses program-controlled aperture to achieve changes in depth of field and create a background blur effect, regardless of the specific content or quality of the background. Furthermore, the system uses edge detection algorithms to enhance the edge contours of special objects and draws a dotted frame around them for identification. The marking operation is independent of the nature of the object or other elements of the image; When a special object needs to be visually enlarged to emphasize its location in the image, the system digitally enlarges the selected area according to the pre-set magnification ratio, and the operation is based only on the original size and location information of the object; if the magnification operation causes the pixel resolution to change, the system will execute a sharpening algorithm to adjust the clarity of the enlarged area, which is independent of the overall quality of the image; If real-time analysis indicates that the image histogram tends to the low-brightness area, indicating that the lighting conditions may lead to underexposure, the system will automatically increase the ISO value and reduce the shutter speed, or adjust the aperture to a larger value; when the histogram tends to the high-brightness area, suggesting that there is a risk of overexposure, the system will reduce the ISO value, increase the shutter speed or select a smaller aperture value to limit the entry of light. If a special object needs to be highlighted in the image, but the background information needs to be kept to provide context, the system will slightly adjust the aperture to a medium size to achieve a slight background blur effect. Once the background blur is complete, if the edge of the special object needs to be enhanced to improve its definition in the image, an edge enhancement filter will be applied and a dotted box will be drawn around the object to mark it. The marking process takes into account the overall style of the image to ensure that it matches the background blur strategy. When a special object needs to be more visually obvious due to its importance, the system digitally enlarges the target object according to the predetermined enlargement ratio, and adjusts the sharpness of the enlarged area to adapt to the resolution change after the enlargement; if the enlargement result causes the details of the special object to be unclear, the details will be sharpened, but over-sharpening will be avoided to avoid introducing artificial noise. The sharpening operation is limited to maintaining the clarity of the enlarged details without changing other parts of the image; If the score is in the high score range, it means that the image analysis shows a slight color temperature deviation, and the system automatically adjusts the white balance parameters based on an accurate assessment of the current image color distribution to match the natural light color temperature; when the image shows a slight exposure deviation, the system fine-tunes the exposure compensation value to cope with subtle lighting changes, keeping the adjustment range to a minimum to avoid significantly changing the overall image exposure effect; If a special object needs to be slightly highlighted in the image, but the background information needs to be kept intact, the aperture size is slightly adjusted to minimize the background blur. When a special object needs to be carefully marked to improve recognition, a non-invasive marking method is used to slightly adjust the brightness or contrast or draw a subtle border to avoid the marking process interfering with the visual content of the image. When the details of a special object are identified to be important for the overall understanding, but it is necessary to avoid disrupting the image balance, slightly enlarge the target object, carefully control the proportions during the enlargement process to ensure that the overall composition of the image is not affected; if the enlargement results in the loss of details of the special object, apply detailed sharpening processing, adjust the enlarged area, and keep other operations unchanged.
6. The method for optimizing the management model of video images shot by a drone as claimed in claim 5, Features: The noise identification algorithm analyzes the image including, the image is converted to the frequency domain to identify the noise: Among them, f(x,y) is the representation of the original image in the spatial domain, and F(u,v) is the representation of the image in the frequency domain. represents Fourier transform; Perform waveform threshold processing. In the frequency domain, identify and process noise according to the set threshold T: Where F′(u,v) is the processed frequency domain representation, in which the frequency components below the threshold T are considered as noise and set to 0; By performing an inverse Fourier transform, the noise-processed frequency domain image is converted back to the spatial domain: in, represents the inverse Fourier transform, and f′(x,y) is the representation of the denoised image in the spatial domain; Apply the Unsharp Mask technique to enhance the edges and details of the image: g(x,y)=f′(x,y)+k·(f′(x,y)-B(x,y)) Among them, g(x,y) is the sharpened image, B(x,y) is the blurred image obtained by Gaussian blurring the original image f′(x,y), and k is the enhancement coefficient used to adjust the intensity of sharpening; An adaptive threshold setting is introduced for waveform threshold processing, and the threshold is dynamically adjusted based on the local frequency domain characteristics of the image: Among them, T adaptive (u,v) is the adaptive threshold at the frequency domain coordinate (u,v), F local (u,v) represents the frequency component in the local frequency domain, std is the standard deviation function, which is used to evaluate the variability in the local frequency domain, and and ρ is the adjustment coefficient; Perform local noise suppression in the frequency domain rather than globally: Noise components are suppressed in the frequency domain based on the ratio of their intensity relative to a local adaptive threshold, allowing for more flexible processing of noise in different frequency domain regions; Use edge detection algorithms to guide the sharpening process so that the sharpening is mainly concentrated in the edge area and the impact on the smooth area is reduced: g′(x,y)=g(x,y)+λ·(g(x,y)-g smooth (x,y))·E(x,y) Among them, g smooth (x, y) is the result of Gaussian blurring of image g(x, y), E(x, y) is the edge intensity map obtained from the original image f′(x, y), and λ is an adjustment factor that controls the adaptive adjustment of sharpening intensity as the edge intensity changes; Through edge-aware sharpening technology, the sharpening effect is enhanced in the edge area of the image and naturally suppressed in the non-edge smooth area, thereby minimizing the artifacts and noise enhancement that may be generated during the sharpening process.
7. The method for optimizing the management model of video images shot by a drone as claimed in claim 6, Features: The dynamically adjusting the flight path using image quality feedback includes: For video images that were once evaluated as low scores, the system analyzes the captured images in real time, determines that the problem is underexposure or overexposure, and adjusts the flight parameters; when the video image is underexposed, the system increases the ISO value, reduces the aperture f-number, and reduces the shutter speed to introduce a large amount of light; if the image quality does not improve significantly, the system will re-plan the flight path, adjust the height of the drone or change the angle to obtain the best lighting conditions or avoid backlight; the system will continue to monitor the image quality and continue to adjust the flight speed and camera parameters as needed until a satisfactory image quality level is achieved; For video images that were once assessed as medium, the system fine-tunes the camera's ISO, aperture, and shutter speed. Considering the importance of shooting angle and timing to image quality, the system fine-tunes the drone's flight path based on real-time feedback to find the optimal shooting position and angle, while also considering the shooting time to take advantage of the best natural lighting conditions, while emphasizing the use of the golden hour's lighting advantages and avoiding backlight or excessive direct light; For video images that have been evaluated as highly rated, the system first ensures flight stability by monitoring flight stability and fine-tuning flight parameters to adapt to subtle changes in ambient lighting. This prevents unnecessary blur or vibration introduced by flight operations. For images that already have high quality, the system accurately monitors slight changes in lighting conditions and adjusts white balance and exposure compensation in a timely manner to maintain color accuracy and the overall brightness balance of the image. It also slightly adjusts flight altitude and speed as needed and optimizes shooting angles to ensure that the best viewing angle and lighting effects are captured.
8. A system using the optimization method of the management model of video images shot by a drone as claimed in any one of claims 1 to 7, Features: The system includes an image acquisition module, an image processing module, a drone control module, a user interface module, a path planning module and a dynamic adjustment module; The image acquisition module is configured with a high dynamic range HDR sensor for acquiring video frames at different exposure levels; The image processing module is equipped with an HDR image synthesis algorithm, a real-time noise suppression algorithm, an automatic exposure adjustment algorithm and a color correction algorithm; wherein the HDR image synthesis algorithm processes video frames of different exposure levels acquired by the image acquisition module to generate HDR video frames; the real-time noise suppression algorithm is used to reduce noise in video frames acquired under low light conditions; the automatic exposure adjustment algorithm is used to automatically adjust exposure parameters according to real-time video analysis; the color correction algorithm is used to optimize the color performance of video frames; The drone control module is equipped with an adaptive flight strategy logic unit and an environment perception unit, which can adjust the flight parameters of the drone according to the analysis results of the image processing module, and use the light sensor data to assist in adjusting the flight and shooting parameters; The user interface module provides an operation interface that allows the user to manually adjust image acquisition parameters, view real-time video previews, and receive parameter adjustments suggested by the system; The path planning module is used to optimize the UAV flight parameters according to the image quality feedback, set the flight path planning parameters according to the target shooting area and the expected image quality requirements, and dynamically adjust the flight path using the image quality feedback; The dynamic adjustment module is used to perform secondary evaluation, automatically adjust camera shooting parameters, and process images based on the content recognized by the model.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program. It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
Citation Information
Cited By
Unmanned aerial vehicle image optimization system and method based on image matching
CN119223294A
Ultrahigh-definition long-distance full-color imaging method under extremely low illumination and module thereof
CN119767153A
Unmanned aerial vehicle continuous frame image splicing method and system
CN120125431A
Low-altitude unmanned aerial vehicle live video real-time rapid mapping method and rapid mapping system
CN120166301A
Unmanned aerial vehicle-based scenic spot real-time data acquisition and processing method
CN120220002A
