A method and system for rapidly processing images of aerial remote sensing imaging equipment

Abstract

The present invention belongs to the technical field of aerial remote sensing and discloses a method and system for quickly processing images of aerial remote sensing imaging equipment. The method comprises: acquiring an aerial remote sensing image, initializing the pixel attribute value of each pixel point therein, determining a road area in the aerial remote sensing image, and determining a plurality of pixel blocks each consisting of a plurality of adjacent pixel points in a portion of the aerial remote sensing image corresponding to the road area; extracting a feature value of each pixel block respectively, and inputting the feature value of each pixel block into a machine learning model in sequence to obtain the probability that each pixel block represents a vehicle, and when the probability is greater than a probability threshold, determining that the corresponding pixel block represents a vehicle; and calculating the density of vehicles and the average speed of vehicles in the portion of the aerial remote sensing image respectively. The present invention solves the technical problem that it is difficult to accurately identify moving vehicles in low-resolution aerial remote sensing images.

Description

Technical Field

[0001] The present invention belongs to the technical field of aerial remote sensing, and in particular relates to a method and system for rapidly processing images of aerial remote sensing imaging equipment. Background Art

[0002] Aerial remote sensing refers to remote sensing technology that uses various aircraft, balloons, etc. as carriers of imaging equipment to conduct aerial remote sensing images in the air. Aerial remote sensing images of objects on the ground can be obtained. Different aerial remote sensing imaging equipment produces aerial remote sensing images with different resolutions. High-resolution aerial remote sensing images often require expensive aerial remote sensing imaging equipment. In the prior art, there are application scenarios for analyzing ground traffic conditions based on aerial remote sensing images of objects on the ground. If high-resolution aerial remote sensing images are used, the analysis cost will be greatly increased. If low-resolution aerial remote sensing images are used, the analysis cost can be reduced. However, in low-resolution aerial remote sensing images, vehicles on the ground appear as pixel blocks composed of multiple adjacent pixels, which are generally difficult to identify. Therefore, the present invention provides a method and system for processing images of aerial remote sensing imaging equipment to analyze ground traffic conditions based on low-resolution aerial remote sensing images. Summary of the Invention

[0003] The present invention addresses the technical problem of difficulty in accurately identifying moving vehicles in low-resolution aerial remote sensing images. The present invention narrows the scope of aerial remote sensing images for identifying moving vehicles by determining the road area in the aerial remote sensing images, and extracts pixel blocks that may represent moving vehicles from the portion of the aerial remote sensing images corresponding to the road area. At the same time, a machine learning model is used to further determine whether the pixel blocks represent moving vehicles. Finally, traffic conditions are analyzed based on the pixel blocks representing vehicles in the portion of the aerial remote sensing images.

[0004] In order to achieve the above-mentioned object of the invention, a method for rapidly processing images of an aerial remote sensing imaging device is provided as follows, which mainly comprises the following steps:

[0005] Acquiring an aerial remote sensing image, initializing pixel attribute values ​​of each pixel therein, the pixel attribute values ​​including a grayscale value and an RGB value of the pixel, extracting a plurality of edge point curves from the aerial remote sensing image, and determining a road area in the aerial remote sensing image based on the plurality of edge point curves;

[0006] In a portion of the aerial remote sensing image corresponding to the road area, a plurality of pixel blocks each consisting of a plurality of adjacent pixel points are determined from the portion of the aerial remote sensing image according to a pixel attribute value of each pixel point therein;

[0007] Extracting feature values ​​of each pixel block, respectively, wherein the feature values ​​of the pixel block include a position of the pixel block in a portion of the aerial remote sensing image corresponding to the road area and a positional relationship between the pixel block and other pixel blocks, and sequentially inputting the feature values ​​of each pixel block into a trained machine learning model to obtain a probability that each pixel block represents a vehicle, and determining that the corresponding pixel block represents a vehicle when the probability is greater than a predetermined probability threshold;

[0008] Based on the pixel blocks representing vehicles, the number of vehicles in the portion of the aerial remote sensing image is counted, and combined with the pixel resolution of the aerial remote sensing image, the density of vehicles and the average speed of vehicles in the portion of the aerial remote sensing image are calculated respectively to obtain the traffic conditions in the portion of the aerial remote sensing image.

[0009] As a preferred technical solution of the present invention, determining the road area in the aerial remote sensing image includes the following steps:

[0010] In the aerial remote sensing image, pixel points having rapidly changing pixel attribute values ​​are determined, and adjacent pixel points having rapidly changing pixel attribute values ​​are connected to form a plurality of edge point curves, one edge point curve is selected from the plurality of edge point curves as a first edge point curve, and other edge point curves are determined that are parallel to the first edge point curve, and one edge point curve is selected from the other edge point curves as a second edge point curve;

[0011] sequentially calculating pixel position differences between corresponding pixels on the first edge point curve and on the second edge point curve, and calculating an average pixel position difference between the first edge point curve and the second edge point curve based on the number of pixels on the shorter of the first and second edge point curves, continuously selecting a next edge point curve from the remaining edge point curves as a new second edge point curve, and calculating the average pixel position difference between the first and second edge point curves;

[0012] The second edge point curve corresponding to the minimum average pixel position difference is paired with the first edge point curve, and the area in the aerial remote sensing image defined by the two is the road area in the aerial remote sensing image. The next edge point curve is continuously selected from the plurality of edge point curves as the new first edge point curve, and the above steps are repeated.

[0013] As a preferred technical solution of the present invention, determining a plurality of pixel blocks each consisting of a plurality of adjacent pixel points from the partial aerial remote sensing image comprises the following steps:

[0014] Based on the partial aerial remote sensing image, two adjacent pixel points having the most similar pixel attribute values ​​are searched therein, and the two adjacent pixel points having the most similar pixel attribute values ​​are merged into a large pixel area, wherein the pixel attribute value of the pixel area is an average of the pixel attribute values ​​of the two adjacent pixel points;

[0015] The pixel area is also regarded as a pixel point, and the previous step is repeated until there are no two adjacent pixel points with the most similar pixel attribute values ​​in the partial aerial remote sensing image, and each pixel area in the partial aerial remote sensing image at this time is regarded as the final pixel block.

[0016] As a preferred technical solution of the present invention, determining a plurality of pixel blocks each consisting of a plurality of adjacent pixel points from the partial aerial remote sensing image further includes the following steps:

[0017] Determining a size of a rectangular pixel window based on the partial aerial remote sensing image, moving the rectangular pixel window in the partial aerial remote sensing image, and calculating an average value of the pixel attribute values ​​of all pixels within the rectangular pixel window each time the rectangular pixel window is moved;

[0018] Acquire at least two other parts of the aerial remote sensing images corresponding to the road area taken at different times, and move the same rectangular pixel window to the same position in each of the other parts of the aerial remote sensing images, calculate the average value of the pixel attribute values ​​of all the pixels within the same rectangular pixel window, and compare it with the average value of the pixel attribute values ​​of all the pixels within the rectangular pixel window in the previous step. When the difference between the two is greater than a preset difference threshold, form a final pixel block with all the pixels within the rectangular pixel window in the previous step.

[0019] As a preferred technical solution of the present invention, the density c of vehicles in the partial aerial remote sensing image is calculated by the following formula 1:

[0020]

[0021] Wherein, n refers to the total number of pixel blocks representing vehicles in the portion of the aerial remote sensing image, k refers to the total number of pixel points in the road area corresponding to the portion of the aerial remote sensing image, and r refers to the pixel resolution of the portion of the aerial remote sensing image in meters / pixel.

[0022] As a preferred technical solution of the present invention, calculating the average speed of the vehicle in the portion of the aerial remote sensing image includes the following steps:

[0023] The partial aerial remote sensing image is used as a first partial aerial remote sensing image, and a second partial aerial remote sensing image is obtained that is taken after the first partial aerial remote sensing image and also corresponds to the road area, and a first pixel block and a second pixel block representing the same vehicle are determined in the first partial aerial remote sensing image and the second partial aerial remote sensing image, respectively.

[0024] The speed v of the vehicle represented by the first pixel block and the second pixel block is calculated using the following formula 2:

[0025] (x1,y1)=∑(p,q) / u, (x2,y2)=∑(j,s) / w

[0026] Where r is the pixel resolution of the partial aerial remote sensing image in meters per pixel, (x2, y2) is the pixel position of the geometric center of the second pixel block, (j, s) is the pixel position of the pixel points in the second pixel block, w is the total number of pixels in the second pixel block, (x1, y1) is the pixel position of the geometric center of the first pixel block, (p, q) is the pixel position of the pixel points in the first pixel block, u is the total number of pixels in the first pixel block, t2 and t1 are the capture times of the second and first partial aerial remote sensing images;

[0027] Calculate the speed of all vehicles in a portion of the aerial remote sensing image, and then calculate the average speed of all vehicles.

[0028] The present invention also provides a system for rapidly processing images from aerial remote sensing imaging equipment, comprising the following modules:

[0029] a preprocessing module for acquiring an aerial remote sensing image, initializing a pixel attribute value of each pixel therein, and determining a road area in the aerial remote sensing image, and determining a plurality of pixel blocks each consisting of a plurality of adjacent pixels in a portion of the aerial remote sensing image corresponding to the road area;

[0030] a detection module, configured to extract a feature value of each pixel block, input the feature value of each pixel block into a trained machine learning model, and obtain a probability that each pixel block represents a vehicle. When the probability is greater than a preset probability threshold, the corresponding pixel block is determined to represent a vehicle.

[0031] The analysis module is configured to count the number of vehicles in the portion of aerial remote sensing imagery, and to calculate the density and average speed of the vehicles in the portion of aerial remote sensing imagery, thereby obtaining traffic information for the portion of aerial remote sensing imagery. Further details on the functions of each module can be found in the method for rapidly processing aerial remote sensing imaging equipment images of the same inventive concept.

[0032] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium is used to store program code, and the program code is used to execute any one of the above methods.

[0033] According to another aspect of the present invention, a computing device is provided, comprising a processor and a memory.

[0034] The memory is used to store program code and transmit the program code to the processor;

[0035] The processor is configured to execute any one of the above methods according to instructions in the program code.

[0036] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0037] 1. The present invention first obtains an aerial remote sensing image, initializes the pixel attribute value of each pixel therein, and determines the road area in the aerial remote sensing image. In the portion of the aerial remote sensing image corresponding to the road area, a number of pixel blocks each consisting of multiple adjacent pixels are determined. Next, the feature value of each pixel block is extracted and input into a machine learning model in sequence to obtain the probability that each pixel block represents a vehicle. When the probability is greater than a probability threshold, the corresponding pixel block is determined to represent a vehicle. Finally, the density of vehicles and the average speed of vehicles in the portion of the aerial remote sensing image corresponding to the road area are calculated to analyze the traffic conditions in the road area.

[0038] 2. The present invention can solve the problem of difficulty in identifying moving vehicles in low-resolution aerial remote sensing images, reduce the analysis cost of analyzing traffic conditions on roads based on aerial remote sensing images, and ensure the accuracy of identifying moving vehicles in low-resolution aerial remote sensing images, thereby obtaining accurate analysis results of traffic conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A flowchart of the steps of a method for rapidly processing images of aerial remote sensing imaging equipment according to the present invention;

[0040] Figure 2 The present invention is a structural diagram of a system for rapidly processing images from aerial remote sensing imaging equipment. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0042] It is understood that the terms "first," "second," etc., used herein may be used to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script without departing from the scope of this application.

[0043] References Figure 1 As shown, the present invention provides a method for quickly processing images of aerial remote sensing imaging equipment, which is mainly achieved by performing the following steps:

[0044] Step 1: Acquire an aerial remote sensing image, initialize pixel attribute values ​​for each pixel therein, wherein the pixel attribute values ​​include a grayscale value and an RGB value of the pixel, extract a plurality of edge point curves from the aerial remote sensing image, and determine a road area in the aerial remote sensing image based on the plurality of edge point curves;

[0045] Furthermore, determining the road area in the above-mentioned aerial remote sensing image includes the following steps:

[0046] The first step is to determine, in the aerial remote sensing image, each pixel point having a rapidly changing pixel attribute value, and connect adjacent pixel points having rapidly changing pixel attribute values ​​to form a plurality of edge point curves, select one edge point curve from the plurality of edge point curves as a first edge point curve, and determine other edge point curves parallel to the first edge point curve, and select one edge point curve from the other edge point curves as a second edge point curve;

[0047] Step 2: Calculate the pixel position difference between corresponding pixels on the first edge point curve and the second edge point curve respectively, and calculate the average pixel position difference between the first edge point curve and the second edge point curve based on the number of pixels on the shorter of the first and second edge point curves. Continuously select the next edge point curve from the other edge point curves as a new second edge point curve, and calculate the average pixel position difference between the first edge point curve and the new second edge point curve.

[0048] Step 3: Pair the second edge point curve corresponding to the minimum average pixel position difference with the first edge point curve. The area in the aerial remote sensing image defined by the second edge point curve and the first edge point curve is the road area in the aerial remote sensing image. The next edge point curve is selected from the plurality of edge point curves as the new first edge point curve, and the above steps are repeated.

[0049] Specifically, the inventors took into account that aerial remote sensing images acquired by aerial remote sensing imaging equipment generally correspond to a larger ground area, including images of roads, buildings, natural landscapes, etc. If a moving vehicle is to be identified in the entire range of aerial remote sensing images, it will not only be difficult, because in low-resolution aerial remote sensing images, the images of moving vehicles or other objects may appear as blurred pixel blocks composed of adjacent pixels, but it will also take more time to identify the moving vehicle in the aerial remote sensing image. In order to avoid these problems, a method for determining the road area in the aerial remote sensing image is proposed: adjacent pixel points with rapidly changing pixel attribute values ​​in the aerial remote sensing image are connected into different edge point curves, and the edge point curves represent the road edges in the aerial remote sensing image. In addition, in different edge point curves, two parallel edge point curves with the smallest average pixel position difference between them are respectively formed into an edge point curve pair. The two edge point curves are respectively the two edges of the road in the aerial remote sensing image. The formula for calculating the average pixel position difference of the two edge point curves is: Among them, (x i ,y i ), (x′ i ,y′ i ) are the pixel positions of the corresponding pixel points on the two edge point curves, and n is the number of pixels of the shorter of the two edge point curves; this method can narrow the range of aerial remote sensing images for identifying moving vehicles, improve the accuracy of identifying moving vehicles, and also improve the efficiency of identifying moving vehicles.

[0050] Step 2: in the portion of the aerial remote sensing image corresponding to the road area, according to the pixel attribute value of each pixel therein, determining a plurality of pixel blocks each consisting of a plurality of adjacent pixel points from the portion of the aerial remote sensing image;

[0051] Furthermore, a plurality of pixel blocks each consisting of a plurality of adjacent pixel points is determined from the aforementioned partial aerial remote sensing image, including the following steps:

[0052] The first step is to find two adjacent pixels with the most similar pixel attribute values ​​in the partial aerial remote sensing image, and merge the two adjacent pixels with the most similar pixel attribute values ​​into a large pixel area, where the pixel attribute value of the pixel area is the average of the pixel attribute values ​​of the two adjacent pixels;

[0053] In the second step, the above pixel area is also regarded as a pixel point, and the above step is repeated until there are no two adjacent pixel points with the most similar pixel attribute values ​​in the above part of the aerial remote sensing image, and each pixel area in the above part of the aerial remote sensing image at this time is regarded as the final pixel block;

[0054] Specifically, based on the partial aerial remote sensing image corresponding to the road area determined in the above step 1, it is only necessary to identify the moving vehicle in the partial aerial remote sensing image, because the moving vehicle appears as a fuzzy pixel block composed of multiple adjacent pixels in the low-resolution aerial remote sensing image. Therefore, in order to identify the moving vehicle, it is necessary to extract the fuzzy pixel block that may represent the vehicle in the partial aerial remote sensing image. First, for each pixel in the partial aerial remote sensing image, find a pixel point with the most similar pixel attribute value. If the two pixel points are adjacent, then merge the two pixel points to form a pixel area, and regard this pixel area as a pixel point, and continue to find another pixel point with the most similar pixel attribute value. If such a pixel point exists, continue to merge the pixels until there are no two adjacent pixels in the partial aerial remote sensing image whose pixel attribute values ​​are the most similar.

[0055] Furthermore, a plurality of pixel blocks each consisting of a plurality of adjacent pixel points can be determined from the aforementioned partial aerial remote sensing image by the following steps:

[0056] The first step is to determine the size of a rectangular pixel window based on the partial aerial remote sensing image, move the rectangular pixel window in the partial aerial remote sensing image, and calculate the average value of the pixel attribute values ​​of all pixels within the rectangular pixel window each time the rectangular pixel window is moved;

[0057] Step 2: Obtain at least two other parts of the aerial remote sensing images corresponding to the road area, taken at different times, and move the same rectangular pixel window to the same position in each of the other parts of the aerial remote sensing images. Calculate the average value of the pixel attribute values ​​of all pixels within the same rectangular pixel window, and compare the average value with the average value of the pixel attribute values ​​of all pixels within the rectangular pixel window in the previous step. When the difference between the two values ​​is greater than a preset difference threshold, form a final pixel block with all pixels within the rectangular pixel window in the previous step.

[0058] Specifically, based on the partial aerial remote sensing image corresponding to the road area determined in the above step 1, the fuzzy pixel blocks that may represent vehicles can also be extracted from the partial aerial remote sensing image by the methods described in the above first and second steps. First, the size of the rectangular pixel window is set according to the total number of pixels in the partial aerial remote sensing image, wherein the length and width of the rectangular pixel window are both integer multiples of the pixels, and the rectangular pixel window is continuously moved in the partial aerial remote sensing image, and the average value of the pixel attribute values ​​of the pixels in the rectangular pixel window is calculated. Secondly, because the vehicle is moving, the position of the same vehicle in the partial aerial remote sensing images taken at different times is different. That is, when the average value of the pixel attribute values ​​of the pixels in the rectangular pixel window at the same position in the partial aerial remote sensing images taken at different times is greatly different, the pixel block composed of the pixels in the rectangular pixel window is more likely to represent the vehicle. The above two methods can achieve the purpose of quickly extracting pixel blocks that may represent vehicles in the partial aerial remote sensing image.

[0059] Step 3: extracting feature values ​​of each pixel block, the feature values ​​of each pixel block including the position of the pixel block in the portion of the aerial remote sensing image corresponding to the road area and the positional relationship between the pixel block and other pixel blocks, and inputting the feature values ​​of each pixel block into a trained machine learning model to obtain a probability that each pixel block represents a vehicle. When the probability is greater than a predetermined probability threshold, the corresponding pixel block is determined to represent a vehicle.

[0060] Specifically, in step 2, different pixel blocks that may represent vehicles have been extracted from some aerial remote sensing images. In order to further improve the accuracy of vehicle recognition, in step 3, the trained machine learning model is continued to be used to judge whether the pixel block represents a vehicle. In this embodiment, there is no restriction on the type of machine learning model. By inputting the characteristic value of the pixel block into the machine learning model, the probability that the pixel block represents a vehicle can be obtained. This is because the characteristic value of the pixel block can be used to determine to a certain extent whether the pixel block represents a vehicle. For example, when the pixel block is too close to the edge of the road area, the pixel block is likely not to represent a vehicle, and when different pixel blocks intersect, the pixel block is likely not to represent a vehicle.

[0061] Step 4: Based on the pixel blocks representing vehicles, the number of vehicles in the portion of the aerial remote sensing image is counted, and the density and average speed of vehicles in the portion of the aerial remote sensing image are calculated in combination with the pixel resolution of the aerial remote sensing image to obtain traffic conditions in the portion of the aerial remote sensing image.

[0062] Furthermore, the density c of vehicles in the above-mentioned aerial remote sensing images is calculated using the following formula 1:

[0063]

[0064] Wherein, n refers to the total number of pixel blocks representing vehicles in the aforementioned partial aerial remote sensing image, k refers to the total number of pixel points in the aforementioned road area corresponding to the aforementioned partial aerial remote sensing image, and r refers to the pixel resolution of the aforementioned partial aerial remote sensing image in meters per pixel. Specifically, after calculating the vehicle density c in the road area corresponding to the aforementioned partial aerial remote sensing image, the relationship between the vehicle density c and a vehicle density threshold is further determined. When the vehicle density c is greater than the vehicle density threshold, it can be considered that the traffic condition in the road area corresponding to the aforementioned partial aerial remote sensing image is congested.

[0065] Furthermore, the average speed of the vehicles in the above-mentioned portion of aerial remote sensing images is calculated, including the following steps:

[0066] The first step is to use the partial aerial remote sensing image as a first partial aerial remote sensing image, and obtain a second partial aerial remote sensing image taken after the first partial aerial remote sensing image and corresponding to the road area, and simultaneously determine a first pixel block and a second pixel block representing the same vehicle in the first partial aerial remote sensing image and the second partial aerial remote sensing image, respectively.

[0067] Step 2: Calculate the speed v of the vehicle represented by the first pixel block and the second pixel block using the following formula 2:

[0068] (x1,y1)=∑(p,q) / u, (x2,y2)=∑(j,s) / w

[0069] Where r is the pixel resolution of the partial aerial remote sensing image in meters per pixel, (x2, y2) is the pixel position of the geometric center of the second pixel block, (j, s) is the pixel position of the pixel points in the second pixel block, w is the total number of pixels in the second pixel block, (x1, y1) is the pixel position of the geometric center of the first pixel block, (p, q) is the pixel position of the pixel points in the first pixel block, u is the total number of pixels in the first pixel block, t2 and t1 are the capture times of the second and first partial aerial remote sensing images;

[0070] Step 3: Calculate the speed of all vehicles in the partial aerial remote sensing image, and then calculate the average speed of all vehicles;

[0071] Specifically, the speed of the vehicle can be calculated using the above formula 2 based on the positions of the pixel blocks representing the same vehicle in some aerial remote sensing images taken at different times. The pixel distance between the first pixel block and the second pixel block is calculated, so the actual moving distance of the vehicle can be calculated. At the same time, the speed of the vehicle is obtained by dividing it by the vehicle's movement time t2-t1, and finally the average speed of all vehicles in the road area corresponding to the partial aerial remote sensing image is calculated; then the average speed of all vehicles and the threshold of the average speed of all vehicles can be compared. When the average speed of all vehicles is less than the threshold of the average speed of all vehicles, it is also considered that the traffic situation in the road area corresponding to the partial aerial remote sensing image is congested.

[0072] References Figure 2 As shown, the present invention also provides a system for processing images of an aerial remote sensing imaging device, which is used to implement the method for processing images of an aerial remote sensing imaging device as described above. Specifically, the functions of each module are described as follows:

[0073] a preprocessing module for acquiring an aerial remote sensing image, initializing a pixel attribute value of each pixel therein, and determining a road area in the aerial remote sensing image, and determining a plurality of pixel blocks each consisting of a plurality of adjacent pixels in a portion of the aerial remote sensing image corresponding to the road area;

[0074] a detection module, configured to extract a feature value of each pixel block, input the feature value of each pixel block into a trained machine learning model, and obtain a probability that each pixel block represents a vehicle. When the probability is greater than a predetermined probability threshold, the corresponding pixel block is determined to represent a vehicle.

[0075] The analysis module is used to count the number of vehicles in the above-mentioned part of the aerial remote sensing image, and calculate the density of vehicles and the average speed of vehicles in the above-mentioned part of the aerial remote sensing image to obtain the traffic situation in the above-mentioned part of the aerial remote sensing image.

[0076] In summary, the present invention first obtains an aerial remote sensing image, initializes the pixel attribute value of each pixel therein, and determines the road area in the aerial remote sensing image, and determines a number of pixel blocks composed of multiple adjacent pixel points in the part of the aerial remote sensing image corresponding to the road area; secondly, the feature value of each pixel block is extracted respectively, and the feature value of each pixel block is input into the machine learning model in sequence to obtain the probability that each pixel block represents a vehicle. When it is greater than the probability threshold, it is determined that the corresponding pixel block represents a vehicle; finally, the density of vehicles and the average speed of vehicles in the part of the aerial remote sensing image corresponding to the road area are calculated respectively to analyze the traffic conditions in the road area; the present invention can solve the problem of difficulty in identifying moving vehicles in low-resolution aerial remote sensing images, reduce the analysis cost of analyzing traffic conditions on roads based on aerial remote sensing images, and ensure the accuracy of identifying moving vehicles in low-resolution aerial remote sensing images, thereby obtaining accurate analysis results of traffic conditions.

[0077] It should be understood that, although the various steps in the flow chart of each embodiment of the present invention are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence according to the order indicated by the arrows. Unless otherwise specified herein, the execution of these steps is not strictly limited in order, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0078] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program. The above-mentioned program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0079] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0080] The above embodiments merely represent several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art may make various modifications and improvements without departing from the scope of the present invention, and these modifications and improvements fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be based on the appended claims.

[0081] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for rapidly processing images from aerial remote sensing imaging equipment, characterized in that: The steps include: Acquiring an aerial remote sensing image, initializing pixel attribute values ​​of each pixel therein, the pixel attribute values ​​including a grayscale value and an RGB value of the pixel, extracting a plurality of edge point curves from the aerial remote sensing image, and determining a road area in the aerial remote sensing image based on the plurality of edge point curves; In a portion of the aerial remote sensing image corresponding to the road area, a plurality of pixel blocks each consisting of a plurality of adjacent pixel points are determined from the portion of the aerial remote sensing image according to a pixel attribute value of each pixel point therein; Extracting feature values ​​of each pixel block, including a position of the pixel block in a portion of the aerial remote sensing image corresponding to the road area and a positional relationship between the pixel block and other pixel blocks, and sequentially inputting the feature values ​​of each pixel block into a trained machine learning model to obtain a probability that each pixel block represents a vehicle. When the probability is greater than a predetermined probability threshold, the corresponding pixel block is determined to represent a vehicle. Based on the pixel blocks representing vehicles, the number of vehicles in the portion of the aerial remote sensing image is counted, and combined with the pixel resolution of the aerial remote sensing image, the density of vehicles and the average speed of vehicles in the portion of the aerial remote sensing image are calculated respectively to obtain the traffic conditions in the portion of the aerial remote sensing image.

2. A method for rapidly processing images from aerial remote sensing imaging equipment according to claim 1, characterized in that: Determining the road area in the aerial remote sensing image includes the following steps: In the aerial remote sensing image, pixel points having rapidly changing pixel attribute values ​​are determined, and adjacent pixel points having rapidly changing pixel attribute values ​​are connected to form a plurality of edge point curves, one edge point curve is selected from the plurality of edge point curves as a first edge point curve, and other edge point curves are determined that are parallel to the first edge point curve, and one edge point curve is selected from the other edge point curves as a second edge point curve; sequentially calculating pixel position differences between corresponding pixels on the first edge point curve and on the second edge point curve, and calculating an average pixel position difference between the first edge point curve and the second edge point curve based on the number of pixels on the shorter of the first and second edge point curves, continuously selecting a next edge point curve from the remaining edge point curves as a new second edge point curve, and calculating the average pixel position difference between the first and second edge point curves; The second edge point curve corresponding to the minimum average pixel position difference is paired with the first edge point curve, and the area in the aerial remote sensing image defined by the two is the road area in the aerial remote sensing image. The next edge point curve is continuously selected from the plurality of edge point curves as the new first edge point curve, and the above steps are repeated.

3. The method for rapidly processing images from aerial remote sensing imaging equipment according to claim 1, characterized in that: Determining a plurality of pixel blocks each consisting of a plurality of adjacent pixel points from the partial aerial remote sensing image comprises the following steps: Based on the partial aerial remote sensing image, two adjacent pixel points having the most similar pixel attribute values ​​are searched therein, and the two adjacent pixel points having the most similar pixel attribute values ​​are merged into a large pixel area, wherein the pixel attribute value of the pixel area is an average of the pixel attribute values ​​of the two adjacent pixel points; The pixel area is also regarded as a pixel point, and the previous step is repeated until there are no two adjacent pixel points with the most similar pixel attribute values ​​in the partial aerial remote sensing image, and each pixel area in the partial aerial remote sensing image at this time is regarded as the final pixel block.

4. The method for rapidly processing images from aerial remote sensing imaging equipment according to claim 1, characterized in that: Determining a plurality of pixel blocks each consisting of a plurality of adjacent pixel points from the partial aerial remote sensing image further includes the following steps: Determining a size of a rectangular pixel window based on the partial aerial remote sensing image, moving the rectangular pixel window in the partial aerial remote sensing image, and calculating an average value of the pixel attribute values ​​of all pixels within the rectangular pixel window each time the rectangular pixel window is moved; Acquire at least two other parts of the aerial remote sensing images corresponding to the road area taken at different times, and move the same rectangular pixel window to the same position in each of the other parts of the aerial remote sensing images, calculate the average value of the pixel attribute values ​​of all the pixels within the same rectangular pixel window, and compare it with the average value of the pixel attribute values ​​of all the pixels within the rectangular pixel window in the previous step. When the difference between the two is greater than a preset difference threshold, form a final pixel block with all the pixels within the rectangular pixel window in the previous step.

5. The method for rapidly processing images of aerial remote sensing imaging equipment according to claim 1, characterized in that: The density c of vehicles in the partial aerial remote sensing image is calculated using the following formula 1: Wherein, n refers to the total number of pixel blocks representing vehicles in the portion of the aerial remote sensing image, k refers to the total number of pixel points in the road area corresponding to the portion of the aerial remote sensing image, and r refers to the pixel resolution of the portion of the aerial remote sensing image in meters / pixel.

6. The method for rapidly processing images of aerial remote sensing imaging equipment according to claim 1, characterized in that: Calculating the average speed of the vehicle in the portion of the aerial remote sensing image comprises the following steps: The partial aerial remote sensing image is used as a first partial aerial remote sensing image, and a second partial aerial remote sensing image is obtained that is taken after the first partial aerial remote sensing image and also corresponds to the road area, and a first pixel block and a second pixel block representing the same vehicle are determined in the first partial aerial remote sensing image and the second partial aerial remote sensing image, respectively. The speed v of the vehicle represented by the first pixel block and the second pixel block is calculated using the following formula 2: Where r is the pixel resolution of the partial aerial remote sensing image in meters per pixel, (x2, y2) is the pixel position of the geometric center of the second pixel block, (j, s) is the pixel position of the pixel points in the second pixel block, w is the total number of pixels in the second pixel block, (x1, y1) is the pixel position of the geometric center of the first pixel block, (p, q) is the pixel position of the pixel points in the first pixel block, u is the total number of pixels in the first pixel block, t2 and t1 are the capture times of the second and first partial aerial remote sensing images; Calculate the speed of all vehicles in a portion of the aerial remote sensing image, and then calculate the average speed of all vehicles.

7. A system for rapidly processing images from aerial remote sensing imaging equipment, for implementing the method according to any one of claims 1 to 6, characterized in that: Includes the following modules: a preprocessing module for acquiring an aerial remote sensing image, initializing a pixel attribute value of each pixel therein, and determining a road area in the aerial remote sensing image, and determining a plurality of pixel blocks each consisting of a plurality of adjacent pixels in a portion of the aerial remote sensing image corresponding to the road area; a detection module, configured to extract a feature value of each pixel block, input the feature value of each pixel block into a trained machine learning model, and obtain a probability that each pixel block represents a vehicle. When the probability is greater than a preset probability threshold, the corresponding pixel block is determined to represent a vehicle. The analysis module is used to count the number of vehicles in the portion of aerial remote sensing images, and respectively calculate the density of vehicles and the average speed of vehicles in the portion of aerial remote sensing images to obtain the traffic conditions in the portion of aerial remote sensing images.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store program code, and the program code is used to execute the method according to any one of claims 1 to 6.

9. A computing device, characterized in that The computing device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the method according to any one of claims 1 to 6 according to instructions in the program code.

Images