Image optimization method for unmanned aerial vehicle imaging technology

Abstract

The invention discloses an image optimization method for an unmanned aerial vehicle imaging technology, and relates to the technical field of sound wave imaging, and the method comprises the following steps: S1, obtaining image information, S2, analyzing the color type of an image, S3, carrying out the preliminary denoising of the image, S4, carrying out the two-round denoising of the image, and S5, recognizing the contour information. Two-dimensional data image information obtained through sound wave imaging processing is transmitted to the image analysis model, pixel points are divided according to color types through the image analysis model, an object and a background can be separated for the first time in the image, the time needed by follow-up image optimization is further shortened, and the image quality is improved. The image information is subjected to two times of denoising processing in different modes, so that the obtained image information is ensured to be higher in quality and clearer, meanwhile, contour information in the image is identified by using a frame selection rectangle, and an inclination angle corresponding to the contour information is obtained, so that an operator can conveniently determine the relative position of a target according to the image information.

Description

Technical Field

[0001] The present invention relates to the field of acoustic wave imaging technology, and in particular to an image optimization method for unmanned aerial vehicle imaging technology. Background Art

[0002] Acoustic imaging technology converts sound signals into visual images, providing strong technical support for the application of drones in operation and maintenance, reconnaissance and other fields. The invention patent with application number CN202320213735.7 discloses "an acoustic imaging device installed on a drone, including an acoustic imager, a drone is provided on the upper end of the acoustic imager, a mounting mechanism is installed on the lower surface of the drone, the mounting mechanism is connected to the acoustic imager, the mounting mechanism includes a fixing mechanism fused to the side surface of the drone and a clamping mechanism clamped in the fixing mechanism, the fixing mechanism includes two symmetrically placed fixing grooves welded to the side surface of the drone and a spring fixedly installed in each of the fixing grooves, and the other end of the spring is connected to a connecting block; by placing the acoustic imager on the lower side of the drone, the drone can be controlled to fly and acoustic imaging operations can be quickly performed on a large area. The acoustic imager and the drone are not welded, and the two can be disassembled for easy replacement and maintenance."

[0003] The above-mentioned existing technology solves the problem that drones cannot collect sound waves during use, but in actual use, the two-dimensional data image information obtained after sound wave imaging processing will be directly output through a visual interface, and the image information is not optimized, resulting in the presented image information containing too much noise. In addition, since the objects and background in the image information are not separated, it is difficult for operators to quickly obtain object detail information and relative positions when observing the image. Summary of the invention

[0004] The purpose of the present invention is to provide an image optimization method for drone imaging technology to solve the problems raised in the above background technology.

[0005] A technical solution to achieve the above purpose is: an image optimization method of unmanned aerial vehicle imaging technology, comprising the following steps:

[0006] S1. Obtaining image information: collecting the acoustic wave information in the detection area, and transmitting the two-dimensional data image information obtained through acoustic wave imaging processing to a storage device;

[0007] S2. Analyze the image color type: Use the image analysis model to perform clustering processing on each pixel in the two-dimensional data image information, classify the types according to the colors of the pixels, calculate the type corresponding to the pixels and the proportion of the type, and directly replace the original pixel value with the obtained value, and transmit the obtained initial two-dimensional data image information to the storage device;

[0008] S3, preliminary image denoising: determine the grayscale value corresponding to each pixel and the grayscale values ​​of the remaining pixels within the set range, and use the similarity number analysis algorithm to calculate the similarity number corresponding to the pixel. If the number is less than the preset value, the current pixel is determined to be noise data, and the grayscale value of the current pixel is replaced with the grayscale value of the similar pixel;

[0009] S4, second round of image denoising: analyzing the pixel values ​​contained in the two-dimensional data image information according to the set window size, obtaining the mean and variance values ​​in the area, calculating the mean and variance values ​​corresponding to the entire two-dimensional data image information according to the mean and variance values ​​of each area and the weight coefficient, processing the current two-dimensional data image information according to the filter coefficient, and deleting the noise data contained in the image;

[0010] S5. Identify contour information: After drawing the contours in the image information, calculate the area of ​​all contours. If the area is less than a threshold, delete the contour information. Determine the inclination angle corresponding to the contour information contained in the current rectangle according to the inclination angle of the selected rectangle.

[0011] Preferably, the step S1 utilizes the acoustic wave sensor carried by the drone to collect acoustic wave information and transmits it to the background data platform, and the background data platform processes the acoustic wave information into two-dimensional data image information and then stores it.

[0012] Preferably, the S2 specifically includes the following steps:

[0013] S201, constructing an image analysis model using a Gaussian probability function, extracting two-dimensional data image information, and transmitting it to the image analysis model;

[0014] S202: clustering each pixel in the two-dimensional data image information using an image analysis model, classifying the pixel into different types according to the color of the pixel, and determining the type to which each pixel belongs.

[0015] Preferably, the step S2 further comprises the following steps:

[0016] S203. After extracting the total number of pixels contained in the image information and the number of pixels corresponding to each color type, count the proportion of pixels of each color type, calculate according to the type to which the pixel belongs and the proportion of the type, and directly replace the original pixel value with the obtained value. After all pixels are processed, the obtained initial two-dimensional data image information is transmitted to the storage device.

[0017] Preferably, the S3 specifically includes the following steps:

[0018] S301, after obtaining the initial two-dimensional data image information, set the range of covered pixels, determine the grayscale value corresponding to each pixel and the grayscale values ​​of the remaining pixels within the set range;

[0019] S302: Calculate the grayscale values ​​of the current pixel and the remaining pixels within a set range using a similarity analysis algorithm to obtain a similarity number corresponding to the pixel.

[0020] Preferably, the S3 further comprises the following steps:

[0021] S303, counting the number of similar pixels to the current pixel, if the number is less than a preset value, the current pixel is determined to be noise data, and the gray value of the current pixel is replaced with the gray value of the similar pixel;

[0022] S304: After processing all the pixels in the initial two-dimensional data image information, the two-dimensional data image information after preliminary denoising is obtained, and the two-dimensional data image information is transmitted to a storage device.

[0023] Preferably, the S4 specifically comprises the following steps:

[0024] S401, obtaining the two-dimensional data image information after initial denoising, selecting a 4×4 window size, analyzing the pixel values ​​contained in the two-dimensional data image information according to the set window size, and calculating the mean and variance values ​​in each window area according to the pixel values ​​in the area;

[0025] S402: Calculate the weight coefficient corresponding to each region using a weight analysis algorithm according to the parameters of the region.

[0026] Preferably, the S4 specifically further comprises the following steps:

[0027] S403, obtaining the mean, variance and weight coefficient of each region, calculating the mean and variance corresponding to the entire two-dimensional data image information based on these data, and using the mean and variance corresponding to the image information to deduce the filter coefficient;

[0028] S404: Process the current two-dimensional image data according to the filter coefficient, delete the noise data contained in the image, obtain the two-dimensional image data after two rounds of denoising, and transmit it to a storage device.

[0029] Preferably, the S5 specifically includes the following steps:

[0030] S501, performing binarization processing on the two-dimensional data image information after the second round of denoising, completely separating the object from the background in the image information, and determining the boundary points corresponding to the object;

[0031] S502, after drawing the contours in the image information, calculate the area of ​​all contours, if the area is less than a threshold, delete the contour information, otherwise keep the contour information;

[0032] S503: Set a selection rectangle, use the selection rectangle to identify all contour information in the two-dimensional data, and determine the inclination angle corresponding to the contour information contained in the current rectangle according to the inclination angle of the selection rectangle.

[0033] Preferably, the S5 further comprises the following steps:

[0034] S504: Output the area and the tilt angle corresponding to the contour in the image information through a visual interface.

[0035] The present invention transmits the two-dimensional data image information obtained through acoustic wave imaging processing to the image analysis model, and uses the image analysis model to divide the pixels according to color type, so that the object and the background can be initially separated in the image, further shortening the time required for subsequent image optimization. In addition, since the image information undergoes two denoising processes in different ways, the obtained image information is guaranteed to be of higher quality and clearer. At the same time, the contour information in the image is identified by using a frame selection rectangle, and the inclination angle corresponding to the contour information is obtained, so that the operator can determine the relative position of the target based on the image information. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 Provide an overall method flow chart for an embodiment of the present invention;

[0037] Figure 2 A flow chart of a method for analyzing image color type provided by an embodiment of the present invention;

[0038] Figure 3 A flow chart of a method for preliminary image denoising provided by an embodiment of the present invention;

[0039] Figure 4 A flow chart of a method for two-round image denoising provided by an embodiment of the present invention;

[0040] Figure 5A flow chart of a method for identifying contour information provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] See also Figure 1-5 The present invention provides a technical solution: an image optimization method of unmanned aerial vehicle imaging technology, comprising the following steps:

[0043] S1. Obtaining image information: using the acoustic wave sensor carried by the drone to collect the acoustic wave information in the detection area, and transmitting the two-dimensional data image information obtained through acoustic wave imaging processing to a storage device;

[0044] S2. Analyze the image color type: Use the image analysis model to perform clustering processing on each pixel in the two-dimensional data image information, classify the types according to the colors of the pixels, calculate the type corresponding to the pixels and the proportion of the type, and directly replace the original pixel value with the obtained value, and transmit the obtained initial two-dimensional data image information to the storage device;

[0045] S3, preliminary image denoising: determine the grayscale value corresponding to each pixel and the grayscale values ​​of the remaining pixels within the set range, and use the similarity number analysis algorithm to calculate the similarity number corresponding to the pixel. If the number is less than the preset value, the current pixel is determined to be noise data, and the grayscale value of the current pixel is replaced with the grayscale value of the similar pixel;

[0046] S4, second round of image denoising: analyzing the pixel values ​​contained in the two-dimensional data image information according to the set window size, obtaining the mean and variance values ​​in the area, calculating the mean and variance values ​​corresponding to the entire two-dimensional data image information according to the mean and variance values ​​of each area and the weight coefficient, processing the current two-dimensional data image information according to the filter coefficient, and deleting the noise data contained in the image;

[0047] S5. Identify contour information: After drawing the contours in the image information, calculate the area of ​​all contours. If the area is less than a threshold, delete the contour information. Determine the inclination angle corresponding to the contour information contained in the current rectangle according to the inclination angle of the selected rectangle.

[0048] S2 specifically includes the following steps:

[0049] S201, constructing an image analysis model using a Gaussian probability function, extracting two-dimensional data image information, and transmitting it to the image analysis model;

[0050] S202, clustering each pixel in the two-dimensional data image information using an image analysis model, classifying the types according to the colors of the pixels, and determining the type to which each pixel belongs. The image analysis model is specifically:

[0051]

[0052] Among them, p(x) represents the probability distribution satisfied by x Gaussian distributions, K represents the number of types, x and k represent parameters, N represents the total number of pixels, and N k (x|μ k ,∑ k ) indicates the probability of the model generating pixel x after the kth model is selected, μ k represents the mean of the model, ∑ k represents the covariance of the Gaussian model, π k represents the prior probability of selecting the k-th model for pixel point x, p(k) represents the probability distribution satisfied by k Gaussian distributions, and p(x|k) represents the probability distribution satisfied by x Gaussian distributions after the k-th model is selected;

[0053] S2 specifically also includes the following steps:

[0054] S203, after extracting the total number of pixels contained in the image information and the number of pixels corresponding to each color type, counting the proportion of pixels of each color type, performing calculations based on the type to which the pixels belong and the proportion of the type, and directly replacing the original pixel value with the obtained value, until all pixels are processed, and then transmitting the obtained initial two-dimensional data image information to a storage device;

[0055] S3 specifically includes the following steps:

[0056] S301, after obtaining the initial two-dimensional data image information, set the range of covered pixels, determine the grayscale value corresponding to each pixel and the grayscale values ​​of the remaining pixels within the set range;

[0057] S302, using a similarity analysis algorithm to calculate the grayscale values ​​of the current pixel and the remaining pixels within a set range to obtain a similarity number corresponding to the pixel. The specific similarity analysis algorithm is:

[0058]

[0059] Among them, L(x) represents the similarity number, I(x,y) represents the similarity result of pixel x and pixel y in the original image, and Ωx represents the domain of pixel x, and y represents the parameter;

[0060] S3 specifically also includes the following steps:

[0061] S303, counting the number of similar pixels to the current pixel, if the number is less than a preset value, the current pixel is determined to be noise data, and the gray value of the current pixel is replaced with the gray value of the similar pixel;

[0062] S304, after processing all the pixels in the initial two-dimensional data image information, obtain the two-dimensional data image information after preliminary denoising, and transmit it to a storage device;

[0063] S4 specifically includes the following steps:

[0064] S401, obtaining the two-dimensional data image information after initial denoising, selecting a 4×4 window size, analyzing the pixel values ​​contained in the two-dimensional data image information according to the set window size, and calculating the mean and variance values ​​in each window area according to the pixel values ​​in the area;

[0065] S402: Calculate the weight coefficient corresponding to each region using a weight analysis algorithm according to the parameters of the region. The weight analysis algorithm is specifically as follows:

[0066]

[0067] Among them, w(i,j,p,q) represents the weight coefficient, K(i,j) represents the normalization constant of the region, ε and a represent parameters, y(i,j) and y(p,q) represent the values ​​corresponding to the pixels in the region, and i,j,p,q represent coordinate parameters;

[0068] S4 specifically further includes the following steps:

[0069] S403, obtaining the mean, variance and weight coefficient of each region, calculating the mean and variance corresponding to the entire two-dimensional data image information based on these data, and using the mean and variance corresponding to the image information to deduce the filter coefficient;

[0070] S404, processing the current two-dimensional data image information according to the filter coefficient, deleting the noise data contained in the image, obtaining the two-dimensional data image information after two rounds of denoising, and transmitting it to a storage device;

[0071] S5 specifically includes the following steps:

[0072] S501, performing binarization processing on the two-dimensional data image information after the second round of denoising, completely separating the object from the background in the image information, and determining the boundary points corresponding to the object;

[0073] S502, after drawing the contours in the image information, calculate the area of ​​all contours, if the area is less than a threshold, delete the contour information, otherwise keep the contour information;

[0074] S503, setting a selection rectangle, using the selection rectangle to identify all contour information in the two-dimensional data, and determining the inclination angle corresponding to the contour information contained in the current rectangle according to the inclination angle of the selection rectangle;

[0075] S504: Output the area and the tilt angle corresponding to the contour in the image information through a visual interface.

[0076] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0077] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An image optimization method for drone imaging technology, characterized in that: The following steps are involved: S1. Obtaining image information: collecting the acoustic wave information in the detection area, and transmitting the two-dimensional data image information obtained through acoustic wave imaging processing to a storage device; S2. Analyze the image color type: Use the image analysis model to perform clustering processing on each pixel in the two-dimensional data image information, classify the types according to the colors of the pixels, calculate the type corresponding to the pixels and the proportion of the type, and directly replace the original pixel value with the obtained value, and transmit the obtained initial two-dimensional data image information to the storage device; S3, preliminary image denoising: determine the grayscale value corresponding to each pixel and the grayscale values ​​of the remaining pixels within the set range, and use the similarity number analysis algorithm to calculate the similarity number corresponding to the pixel. If the number is less than the preset value, the current pixel is determined to be noise data, and the grayscale value of the current pixel is replaced with the grayscale value of the similar pixel; S4, second round of image denoising: analyzing the pixel values ​​contained in the two-dimensional data image information according to the set window size, obtaining the mean and variance values ​​in the area, calculating the mean and variance values ​​corresponding to the entire two-dimensional data image information according to the mean and variance values ​​of each area and the weight coefficient, processing the current two-dimensional data image information according to the filter coefficient, and deleting the noise data contained in the image; S5. Identify contour information: After drawing the contours in the image information, calculate the area of ​​all contours. If the area is less than a threshold, delete the contour information. Determine the inclination angle corresponding to the contour information contained in the current rectangle according to the inclination angle of the selected rectangle.

2. The image optimization method of the drone imaging technology according to claim 1, characterized in that: The step S1 uses the acoustic wave sensor carried by the drone to collect acoustic wave information and transmit it to the background data platform, and the background data platform processes the acoustic wave information into two-dimensional data image information and then stores it.

3. The image optimization method of the drone imaging technology according to claim 1, characterized in that: The S2 specifically includes the following steps: S201, constructing an image analysis model using a Gaussian probability function, extracting two-dimensional data image information, and transmitting it to the image analysis model; S202: clustering each pixel in the two-dimensional data image information using an image analysis model, classifying the pixel into types according to the color of the pixel, and determining the type to which each pixel belongs.

4. The image optimization method of the drone imaging technology according to claim 3 is characterized by: The S2 specifically further includes the following steps: S203. After extracting the total number of pixels contained in the image information and the number of pixels corresponding to each color type, count the proportion of pixels of each color type, calculate according to the type to which the pixel belongs and the proportion of the type, and directly replace the original pixel value with the obtained value. After all pixels are processed, the obtained initial two-dimensional data image information is transmitted to the storage device.

5. The image optimization method of the UAV imaging technology according to claim 1, characterized in that: The S3 specifically includes the following steps: S301, after obtaining the initial two-dimensional data image information, set the range of covered pixels, determine the grayscale value corresponding to each pixel and the grayscale values ​​of the remaining pixels within the set range; S302: Calculate the grayscale values ​​of the current pixel and the remaining pixels within a set range using a similarity analysis algorithm to obtain a similarity number corresponding to the pixel.

6. The image optimization method of the drone imaging technology according to claim 5, characterized in that: The S3 specifically further includes the following steps: S303, counting the number of similar pixels to the current pixel, if the number is less than a preset value, the current pixel is determined to be noise data, and the gray value of the current pixel is replaced with the gray value of the similar pixel; S304: After processing all the pixels in the initial two-dimensional data image information, the two-dimensional data image information after preliminary denoising is obtained, and the two-dimensional data image information is transmitted to a storage device.

7. The image optimization method of the drone imaging technology according to claim 1, characterized in that: The S4 specifically comprises the following steps: S401, obtaining the two-dimensional data image information after initial denoising, selecting a 4×4 window size, analyzing the pixel values ​​contained in the two-dimensional data image information according to the set window size, and calculating the mean and variance values ​​in each window area according to the pixel values ​​in the area; S402: Calculate the weight coefficient corresponding to each region using a weight analysis algorithm according to the parameters of the region.

8. The image optimization method of the drone imaging technology according to claim 7, characterized in that: The S4 specifically further comprises the following steps: S403, obtaining the mean, variance and weight coefficient of each region, calculating the mean and variance corresponding to the entire two-dimensional data image information based on these data, and using the mean and variance corresponding to the image information to deduce the filter coefficient; S404: Process the current two-dimensional image data according to the filter coefficient, delete the noise data contained in the image, obtain the two-dimensional image data after two rounds of denoising, and transmit it to a storage device.

9. The image optimization method of the drone imaging technology according to claim 1, characterized in that: The S5 specifically includes the following steps: S501, performing binarization processing on the two-dimensional data image information after the second round of denoising, completely separating the object from the background in the image information, and determining the boundary points corresponding to the object; S502, after drawing the contours in the image information, calculate the area of ​​all contours, if the area is less than a threshold, delete the contour information, otherwise keep the contour information; S503: Set a selection rectangle, use the selection rectangle to identify all contour information in the two-dimensional data, and determine the inclination angle corresponding to the contour information contained in the current rectangle according to the inclination angle of the selection rectangle.

10. The image optimization method of drone imaging technology according to claim 1, characterized in that: The S5 specifically further comprises the following steps: S504: Output the area and the tilt angle corresponding to the contour in the image information through a visual interface.

Images