Method for fast and safe evaluation of extraterrestrial body surface and landing site optimization

Abstract

The application discloses a method for quickly evaluating a safe area on the surface of an extraterrestrial object by using a single image feature, and belongs to the field of deep space exploration. The method comprises the following steps: collecting a single image of the surface of the extraterrestrial object, removing image noise and rocks smaller than a given size in the image by using a side window filtering method, retaining other rocks, and using a gradient sharpening method to strengthen edge information to obtain a preprocessed image; constructing a ruggedness discrimination operator based on local pixel statistical characteristics, performing image segmentation on the preprocessed image by using the ruggedness discrimination operator, evaluating a safe area, and obtaining a screened binary image of the safe area on the surface of the extraterrestrial object by combining a morphological filtering method; calculating the size of a landing area by using a safety radius spiral search algorithm, combining a calculated value of a ruggedness discrimination function of the safe area, constructing a landing point weighted evaluation method which comprehensively considers the flatness of a landing point and the size of the landing area, and realizing optimization of the landing point. The application can improve the accuracy and efficiency of the safe area and the optimized landing point.

Description

Technical Field

[0001] This invention relates to a method for rapid safety assessment of the surface of extraterrestrial objects and selection of landing sites, and particularly to a method for rapid safety assessment of surface areas and selection of landing sites based on a single image of an extraterrestrial object, belonging to the field of deep space exploration. Background Technology

[0002] Extraterrestrial object exploration is one of the most challenging and technologically complex fields. Among its applications, utilizing small spacecraft, such as CubeSats, for extraterrestrial object exploration is a crucial direction for future deep space exploration development, aligning with the aerospace industry's requirements for low-cost, large-scale development. In exploration missions, due to long communication times and the complex dynamic environment of deep space, ground-based remote control navigation and control methods are no longer sufficient to meet real-time and high-precision requirements; therefore, probes must possess autonomous navigation and control capabilities. Among these, the extraction and recognition of extraterrestrial object surface features is a key technology for achieving spacecraft autonomy. With breakthroughs in computer hardware technology and the development of optical sensing devices, autonomous star surface terrain texture feature recognition and extraction technology based on onboard computers and optical navigation cameras has become a research hotspot.

[0003] As the probe approaches an extraterrestrial object, it needs to conduct real-time safety assessments of the target object's surface, search for potential landing areas, and determine the optimal landing site. Extraterrestrial objects possess numerous craters, rocks, textures, and other topographical features. Under sufficient lighting conditions, optical cameras can capture relatively clear images of the object's surface, providing a wealth of valuable information for safety assessments and landing site selection. Therefore, achieving rapid safety assessments of the target object's surface and optimal landing site selection based on single images is crucial for the operation of the probe's autonomous system.

[0004] Among the developed methods for selecting flat safe zones and landing points, the prior art [1] (Cheng Y, Johnson AE, Matthias LH, et al. Passive imaging based hazard avoidance for spacecraft safe landing [C]. International Symposium on Artificial Intelligence and Robotics & Automation in Space: i-SAIRAS. 2001.) proposed a landing point selection method based on the local gray-scale standard deviation method. By calculating the mean and standard deviation of gray-scale values ​​of each local window in the same image, the point with the smallest local standard deviation is selected as the pre-selected landing point, and the selection results are similar for images of different resolutions. The operation effect of this method in a small neighborhood is similar to edge detection, but because it operates in a local neighborhood, it has a certain resistance to global illumination changes. However, this method only selects the landing point and does not segment the safe zone.

[0005] Prior technology [2] (Woicke, S. and Mooij E. Stereo-Vision Algorithm for Hazard Detection during Planetary Landings [C]. AIAA Guidance, Navigation, and Control Conference, 2014-0272, 2014.) proposed a stereo vision saliency detection method for landers. After effectively detecting obvious protrusions and depressions on the ground, the roughness of the shooting area is evaluated by fuzzy logic decision-making, which improves the accuracy of flat safety area assessment. However, this method uses stereo vision technology as the algorithm basis and requires stereo imaging equipment such as binocular cameras, which is difficult to apply under the large-scale geometry formed by the detector and the target celestial body.

[0006] Existing methods for safety assessment of extraterrestrial surface areas and selection of landing sites do not adequately achieve rapid safety assessment and complete segmentation of potential landing areas, resulting in insufficient extraction of landing area information, large computational load, and slow processing speed for optimal landing site selection. Summary of the Invention

[0007] The purpose of this invention is to provide a method for rapid safety assessment of extraterrestrial surfaces and optimal landing sites. This method constructs a mechanism for rapid assessment of safe areas on celestial surfaces using features from a single image. It introduces a side-window filtering method to eliminate image noise and rocks smaller than a given size, achieving edge-preserving filtering of extraterrestrial surface images. A ruggedness discrimination operator based on local pixel statistical characteristics is constructed, using regional flatness as a measure of safety assessment. This operator performs image segmentation on the preprocessed image, assesses safe areas, and uses morphological filtering to obtain a binarized image of the selected safe areas on the extraterrestrial surface. Finally, a safety radius spiral search algorithm is used to calculate the landing area size, and a weighted landing site assessment method that comprehensively considers both landing site flatness and landing area size is constructed to achieve optimal landing site selection.

[0008] The present invention is achieved through the following technical solution.

[0009] The method for rapid safety assessment of extraterrestrial surfaces and selection of landing sites disclosed in this invention includes the following steps:

[0010] Step 1: Use the side window filtering method to remove image noise and rocks smaller than a given size from the image, retain other rocks, and use the gradient sharpening method to enhance edge information to obtain the preprocessed image.

[0011] Step 1.1: Calculate the filter kernel radius of the side window filter. The method for determining the filter kernel radius is as follows:

[0012]

[0013] Where r is the radius of the filter kernel, and max{a,b} represents the maximum value of elements a and b. This means rounding element A up. This indicates rounding down element B; the horizontal pixel size Δu and the vertical pixel size Δv of the window are represented as follows:

[0014]

[0015] Where f is the focal length of the navigation camera, pe x With PE y Let ΔX and ΔY be the pixel size of the navigation camera, ΔX and ΔY be the given dimensions of the rock in the optical navigation camera coordinate system, and Z be the position of the rock in the optical navigation camera coordinate system. c z c Axial projection;

[0016] The coordinate system of an optical navigation camera is defined as follows: the origin O of the optical navigation camera coordinate system is... c Located at the optical center of the navigation camera, O c z c The axis points in the direction of the navigation camera's optical axis, perpendicular to the image plane, Oc x c Shaft and O c y c The plane formed by the axes is parallel to the image plane and satisfies the right-hand rule.

[0017] Step 1.2: Perform side-window filtering on the image acquired by the optical camera. The side-window filtering method achieves filtering by integrating the filtering results of different sub-filter kernels. The specific filtering method of the sub-filter kernels is as follows:

[0018]

[0019] Among them: I' n U(m) is the filtered output value of the sub-filter kernel. ij I ij ,θ n ,ρ n (r) is the filtering kernel function; I ij To learn how to filter, we need to find the pixel value of point j in the sub-filter kernel of the image acquired by the camera, where point i is the point to be filtered; m ij M represents the filtering weight of point j in the sub-filter kernel with point i as the point to be filtered; n Let m be the sum of the filtering weights for point i to be filtered; i n Let S be the set of points inside the sub-filter kernel; let S be the set of sub-filter kernels; θ n The kernel function parameter is defined as the angle between the sub-filter kernel and the horizontal line; ρ n The kernel function parameter is defined as the length corresponding to the sub-filter kernel;

[0020] By comparing the filtered values ​​I' after processing by each filter kernel n The original pixel value I of the point to be filtered i i The filtering result I' with the minimum Euclidean distance is obtained. SWF The final output of the side window filtering algorithm:

[0021]

[0022] The result of equation (4) achieves the elimination of image noise and rocks smaller than a given size;

[0023] Step 1.3: Use gradient sharpening to process the image, enhancing edge information while preserving features of flat areas; to obtain the preprocessed image.

[0024] Step 2: Construct a ruggedness discrimination operator to segment the preprocessed image obtained in Step 1, evaluate the safe region, and combine it with morphological filtering to obtain the binarized image of the selected safe region.

[0025] Step 2.1: Use the Otsu's method to determine the background of the image; if the pixel is determined to be the background, no ruggedness determination is performed; if the pixel is not determined to be the background, ruggedness determination is performed.

[0026] The optimal threshold for the Otsu's method is T. * Represented as:

[0027]

[0028] in:

[0029]

[0030] In the above formula, n is the total number of pixels in the image, n i M represents the number of pixels corresponding to a certain grayscale value, where M is the maximum grayscale value of the image.

[0031] If the pixel values ​​in the image after preprocessing in step 1 are less than T * Then it is the background, greater than T * Then it is not the background;

[0032] Step 2.2: Achieve rapid assessment of regional security level using the ruggedness discrimination operator. The discrimination function of the ruggedness discrimination operator is:

[0033]

[0034] Where: F(p,q) is the calculated value of the discrimination function of the ruggedness discrimination operator, parameter (p,q) is the coordinate value of the pixel to be estimated, parameter (p+k,q+l) is the coordinate value of the pixels surrounding the pixel to be estimated, and -r≤k≤r, -r≤l≤r, A is the set of interior points of the operator. The difference ΔI between the pixel value I(p+k,q+l) surrounding the pixel to be estimated and the pixel value I(p,q) to be estimated is... pqkl for:

[0035] △I pqkl =I(p,q)-I(p+k,q+l) (8)

[0036] therefore:

[0037]

[0038] In the formula, R kl is the discriminant in the discriminant function; v is the pixel value difference threshold, which is given in advance by the ground.

[0039] The image is binarized using the obtained F(p,q).

[0040]

[0041] In the formula, K is the ruggedness threshold, which is given in advance by the ground. According to the calculation result of formula (10), the safe area in the preprocessed image is quickly segmented to obtain the initial binary image g(p,q).

[0042] Step 2.3: Perform morphological filtering on the initial binarized image obtained in Step 2.2 to eliminate small safe regions and spikes, resulting in a binarized image of the safe region.

[0043] Step 3: By calculating the landing area size SA(p,q) and the safe area F(p,q) obtained in Step 2, a weighted evaluation method for landing points is constructed to evaluate all internal points of the safe area, thereby achieving optimal selection of landing points within the safe area.

[0044] Step 3.1: Use F(p,q) obtained in Step 2 to describe the flatness of the landing area.

[0045] Step 3.2: Use the binarized image of the safe area obtained in Step 2. The landing area size is calculated using the safety radius spiral search algorithm, and the landing area size corresponding to the point within the safety area is SA(p,q).

[0046] SA(p,q) is obtained from the following formula:

[0047] SA(p,q)=πh(p,q)·m(p,q) (11)

[0048] In the above formula, h(p,q) is the safe radius length of the landing area in the x-direction corresponding to pixel coordinate (p,q), and m(p,q) is the safe radius length of the landing area in the y-direction.

[0049] Step 3.3: Evaluate all in-field points within the safe zone using F(p,q) and SA(p,q) to construct a weighted evaluation method for landing points and optimize the landing point selection within the safe zone.

[0050] The landing point evaluation function is:

[0051]

[0052] In the above formula, α represents the weight coefficient of the evaluation function. and This is the normalized expression for the function, with the following specific meanings:

[0053]

[0054] Among them, SA max With F max SA represents the maximum value of the function. min With Fmin This represents the minimum value of the function.

[0055] Therefore, the coordinates corresponding to the maximum value point calculated by equation (12) are the optimal landing point coordinates (p, q) within the safe zone. * :

[0056]

[0057] Among them, H S This is the set of points within the selected safe area.

[0058] This completes the rapid safety assessment of the extraterrestrial body surface and the selection of the optimal landing site.

[0059] Beneficial effects:

[0060] 1. The rapid safety assessment and landing site selection method for extraterrestrial celestial bodies disclosed in this invention introduces a side window filtering method, which can remove image noise and rocks smaller than a given size, realize edge-preserving filtering of extraterrestrial celestial body surface images, and retain more useful information in the images.

[0061] 2. The rapid safety assessment and landing site selection method for the surface of extraterrestrial objects disclosed in this invention constructs a ruggedness discrimination operator to rapidly assess the safety of the surface region of extraterrestrial objects. By calculating the ruggedness discrimination function of the surface region of the celestial body, a threshold discrimination is performed to realize the segmentation of the safe area of ​​the celestial body surface, providing a basis for subsequent landing site selection.

[0062] 3. The rapid safety assessment and landing site selection method for extraterrestrial surfaces disclosed in this invention utilizes a safety radius spiral search algorithm to calculate the landing area of ​​landing sites within the safety region. It then comprehensively considers the ruggedness discrimination function calculated above and the landing area to construct a weighted assessment method for landing sites, obtaining the optimal landing sites within the safety region. This invention improves the accuracy and efficiency of safety region assessment and landing site selection, providing more information for subsequent deep space probe missions. Attached Figure Description

[0063] Figure 1 This is a schematic diagram of the method for rapid safety assessment of extraterrestrial surfaces and selection of landing sites according to the present invention;

[0064] Figure 2 This is a schematic diagram of the side window filtering template in this invention;

[0065] Figure 3 This is a schematic diagram of the image processing effect in step 1 of this embodiment of the invention, using a real-life image from the Hayabusa1 mission as an example. Figure 3 (a) To read the original image, Figure 3 (b) is the image processed by side window filtering. Figure 3(c) is the gradient-sharpened image;

[0066] Figure 4 This is a schematic diagram of the star table ruggedness discrimination result in step 2 of the embodiment of the present invention, wherein... Figure 4 (a) The result of safe zone segmentation using the ruggedness discrimination operator. Figure 4 (b) is the safe zone on the surface of the target celestial body obtained by the open operation process;

[0067] Figure 5 This is a schematic diagram of the method for searching the safe area corresponding to the landing point using the safety radius spiral search algorithm in step 3 of this invention;

[0068] Figure 6 This is a flowchart of the landing point weighted evaluation method in this invention;

[0069] Figure 7 This is a diagram showing the landing point selection results in step 3 of this embodiment and the landing point selection results of the prior art method. Detailed Implementation

[0070] To better illustrate the purpose and advantages of the present invention, the invention will be further described below in conjunction with the accompanying drawings and examples.

[0071] To verify the feasibility of this invention, optical images of the surface of the extraterrestrial object Itokawa were captured, and the safe zone and preferred landing point obtained from image processing of the target object's surface were plotted on the map, as shown below. Figure 4 (b) As shown in Figure 7, the initial position of the detector in the target celestial body B plane coordinate system is [-500; -20; 20] kilometers. The field of view of the optical navigation camera is 0.3 degrees, and the focal length is f = 4300 mm. Mathematical simulation verification is performed.

[0072] like Figure 1 As shown in the figure, the specific implementation steps of the rapid safety assessment and landing site selection method for extraterrestrial celestial bodies disclosed in this embodiment are as follows:

[0073] Step 1: Use the side window filtering method to remove image noise and rocks smaller than a given size, retain other rocks, and use the gradient sharpening method to enhance edge information to obtain the preprocessed image.

[0074] After an optical camera captures an image of the target celestial body's surface, image filtering is used to remove image noise and rocks smaller than a given size, making flat areas more prominent while retaining larger rocks and other rugged terrain information. This embodiment of the invention employs a side-window filtering method, using the same kernel function as mean filtering, to retain edge information of rocks of a certain size while removing image noise and smaller rocks. The original image is shown below. Figure 3 As shown in (a), its pixel size is 612×679 pixels.

[0075] To preserve specific information, the appropriate filter kernel size must first be determined. The size of the filter kernel is related to the detector's height and the camera's field of view. The radius of the filter kernel can be determined using the detector's current relative navigation information.

[0076]

[0077] Where △u and △v are the horizontal and vertical pixel sizes of the window, f is the focal length of the navigation camera, and pe x With PE y Let △X and △Y be the minimum rock size to be determined in the optical navigation camera coordinate system, and Z be the position of the rock in the optical navigation camera coordinate system. c z c Axial projection. The method for converting the focal length of a navigation camera to its field of view is as follows:

[0078]

[0079] Where θ is the size of the field of view, and L is the sensor length corresponding to the field of view.

[0080] Therefore, the filter kernel radius is determined by the following formula:

[0081]

[0082] Where r is the radius of the filter kernel, and max{a,b} represents the maximum value of elements a and b. This means rounding element A up. This indicates rounding down element B. In this embodiment, the minimum rock size is set to 5m, and the filter kernel radius is determined to be 3 pixels through calculation.

[0083] Subsequently, the target celestial image is filtered. In the side-window filtering method, the specific filtering method of the sub-filter kernel is as follows:

[0084]

[0085] Among them: I' n U(m) is the filtered output value of the sub-filter kernel. ij I ij ,θ n ,ρ n (r) is the filtering kernel function; I ij To learn how to filter, we need to find the pixel value of point j in the sub-filter kernel of the image acquired by the camera, where point i is the point to be filtered; m ij M represents the filtering weight of point j in the sub-filter kernel with point i as the point to be filtered; n Let m be the sum of the filtering weights for point i to be filtered; in Let S be the set of points inside the sub-filter kernel; let S be the set of sub-filter kernels; θ n The kernel function parameter is defined as the angle between the sub-filter kernel and the horizontal line; ρ n The kernel function parameter is defined as the length corresponding to the sub-filter kernel.

[0086] In the case of discrete parameters, eight side-window filter sub-kernels are defined, and the parameters corresponding to the eight sub-kernels are as follows: The eight combinations with ρ∈{0,r} Figure 2 The results for the sub-filter kernels (left / right (L, R), up / down (D, U), southwest (SW), southeast (SE), northwest (NW), and northeast (NE)) with a kernel radius r of 2 are given. After filtering using eight sub-filter kernels, the filtered result with the smallest Euclidean distance is obtained by comparing with the original pixel values ​​and is taken as the output of the filter kernel. The filter input is as follows:

[0087]

[0088] Meanwhile, the mean filtering kernel function used in this embodiment is:

[0089]

[0090] Among them, a p,q Let I(p,q) be the pixel value at point (p,q), and u(i,j) be the output of the mean filter kernel function. In the mean filtering method, the weight of each pixel is 1. Figure 3 As shown in (b).

[0091] After performing side-window filtering, to enhance information in non-flat regions and the edges of stones, gradient sharpening is used to process the image, resulting in an image with enhanced edge information while preserving features of flat regions. The gradient sharpening method is as follows:

[0092]

[0093] Among them, G x (p,q) represents the gradient value of the image in the x-direction, G y (p,q) represents the gradient value of the image in the y-direction, and G(p,q) represents the global gradient value at point (p,q). The gradient value G(p,q) is superimposed with the filtering result to obtain the sharpened image H. The final sharpened image is shown below. Figure 3 As shown in (c).

[0094] Step 2: Construct a ruggedness discrimination operator to segment the preprocessed image obtained in Step 1, evaluate the safe region, and combine it with morphological filtering to obtain the binarized image of the selected safe region.

[0095] In this step, a ruggedness discrimination operator is constructed, and the ruggedness discrimination function is used to calculate the value to evaluate the flatness of the area around the current pixel. Combined with morphological filtering methods, a segmented safe region binarized image is obtained.

[0096] First, background discrimination is performed. Since images from deep space environments mainly consist of background and target celestial objects, and the image grayscale histogram exhibits a typical bimodal structure, the maximum inter-class variance (MOV) method is used as the background thresholding method to determine whether a pixel is part of the background. If a pixel is determined to be part of the background, no further background discrimination is performed. Let the optimal threshold be T. * The solution method is as follows:

[0097]

[0098] in:

[0099]

[0100] In the above formula, n is the total number of pixels in the image, n i M represents the number of pixels corresponding to a certain grayscale value, and M is the maximum grayscale value of the image.

[0101] Secondly, the ruggedness discrimination operator calculates whether the difference between the pixel value and the pixel value surrounding the pixel to be estimated exceeds a set pixel value difference threshold. It then determines whether this proportion exceeds the set ruggedness threshold by statistically analyzing the percentage of pixels exceeding the threshold within the entire operator region. If the proportion exceeds the threshold, the pixel is considered to be located in a rugged region; otherwise, it is considered to be located in a flat region. The ruggedness discrimination function is as follows:

[0102]

[0103] Where: F(p,q) is the calculated value of the discrimination function of the ruggedness discrimination operator, parameter (p,q) is the coordinate value of the pixel to be estimated, parameter (p+k,q+l) is the coordinate value of the pixels surrounding the pixel to be estimated, and -r≤k≤r, -r≤l≤r, A is the set of interior points of the operator; ΔI is the difference between the pixel value I(p+k,q+l) surrounding the pixel to be estimated and the pixel value I(p,q). pqkl for:

[0104] △I pqkl =I(p,q)-I(p+k,q+l) (25)

[0105] therefore:

[0106]

[0107] In the formula, R klis the discriminant in the discriminant function, and v is the pixel value difference threshold. In this embodiment, the threshold is estimated by the grayscale radiation value formula and is set to 10.

[0108] The value F(p,q) is calculated using the obtained ruggedness discrimination function, and the image is binarized based on this value.

[0109]

[0110] In the above formula, K is the ruggedness threshold, which is given as 0.85 in this embodiment.

[0111] The obtained ruggedness discrimination image is as follows Figure 4 As shown in (a).

[0112] Finally, to further process the obtained binarized image and eliminate small flat areas and spikes, morphological filtering is performed on the initially selected binarized image. This invention employs an opening operation, with the convolution kernel size consistent with the filter kernel radius obtained by the aforementioned calculation method, i.e., 3 pixels, and the template is a symmetrical elliptical convolution kernel.

[0113] The binarized region obtained through the opening operation is the safe region on the surface of extraterrestrial objects obtained by the fast evaluation method. Figure 4 (b) shows the projection of the final selected safe area onto the original image, as shown in the red area.

[0114] Step 3: By calculating the landing area size SA(p,q) and the safe area F(p,q) obtained in Step 2, a weighted evaluation method for landing points is constructed to evaluate all internal points of the safe area, thereby achieving optimal selection of landing points within the safe area.

[0115] After obtaining the processed binarized image, landing points within the safe area are optimized. Outer contour recognition is performed on all selectable regions in the image, and the area and position enclosed by each contour are calculated for the next step of feature region fitting. In this embodiment, a 4-connectivity method is used to segment each contour. For each contour region, a suitable safe region that meets the scale requirements is generally selected as the target for the next processing step. This embodiment uses the largest selected safe region as an example to optimize the landing points within the safe region.

[0116] The flatness of the landing site area can be described using the ruggedness discrimination function F(p,q) obtained in step two. The size of the landing area is calculated using a safe radius spiral search algorithm. This algorithm searches clockwise from the center of the lander until the search path encounters an obstacle area, at which point the search ends. The area of ​​the ellipse calculated using the safe radius lengths of the map grid covered by the path in both the x and y directions is taken as the size of the landing area corresponding to that landing point. The algorithm principle is as follows: Figure 5 As shown. Let SA(p,q) be the landing area size calculated by the safety radius spiral search algorithm, obtained from the following formula:

[0117] SA(p,q)=πh(p,q)·m(p,q) (28)

[0118] In the above formula, h(p,q) is the safe radius length of the landing area in the x-direction corresponding to pixel coordinate (p,q), and m(p,q) is the safe radius length of the landing area in the y-direction.

[0119] Landing site selection comprehensively considers the flatness of the landing area and the corresponding size of the landing area. Therefore, a weighted evaluation method for landing sites can be designed for evaluation and selection. The principle is as follows: Figure 6 As shown.

[0120] By employing the weighted evaluation method that considers both the flatness and size of the landing area, all points within the safe zone can be evaluated, enabling optimal landing point selection within the safe zone. The weighted evaluation function for landing points is:

[0121]

[0122] In the above formula, α represents the weight coefficient of the evaluation function. and This is the normalized expression for the function, with the following specific meanings:

[0123]

[0124] Among them, SA max With F max SA represents the maximum value of the function. min With F min This represents the minimum value of the function.

[0125] Therefore, the coordinates corresponding to the maximum value point calculated by the evaluation function are the optimal landing point coordinates (p, q). * :

[0126]

[0127] Among them, H S This is the set of points within the selected safe area.

[0128] Optimal results are as follows Figure 7 As shown. This completes the rapid safety assessment of the extraterrestrial body surface and the selection of the optimal landing site. To demonstrate the speed of this method, a comparative simulation is set up to compare it with existing methods that use the local grayscale standard deviation method as a discrimination index for safety area assessment and landing site selection.

[0129] This comparative example compares the method for selecting landing points using local grayscale standard deviation used in the prior art [1] with the method designed in this invention. The main idea of ​​the prior art method is as follows: First, calculate the local grayscale standard deviation value corresponding to each pixel; second, divide the image into nine regions on average, and select the point with the smallest local grayscale standard deviation in each region, thereby obtaining the optimal landing point. This prior art method does not perform evaluation and segmentation of the surface safety area, but only uses the local grayscale standard deviation value to evaluate and select landing points for the global image.

[0130] The local grayscale standard deviation is calculated by the following formula:

[0131]

[0132] Where I(p,q) is the pixel value located at coordinates (p,q) within the window, (i,j) is the coordinates of the window center, w is the window radius, and ISD(i,j) is the local grayscale standard deviation located at coordinates (i,j).

[0133] The results of five separate tests are shown in Table 1 below. Figure 7 As shown in the figure. The results demonstrate that the method proposed in this invention can effectively improve the speed of landing area extraction and landing point selection, and the landing point extraction results of this invention are significantly far away from landing area obstacles, which can better meet landing requirements and has high feasibility.

[0134] Table 1 Comparison of calculations between the method of the present invention and the comparative method.

[0135]

[0136] The above detailed description further illustrates the purpose, technical solution, and beneficial effects of the invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for rapid safety assessment of the surface of an extraterrestrial body and selection of optimal landing sites, characterized in that: Includes the following steps, Step 1: Use the side window filtering method to remove image noise and rocks smaller than a given size from the image, retain other rocks, and use the gradient sharpening method to enhance edge information to obtain the preprocessed image; Step 2: Construct a ruggedness discrimination operator to segment the preprocessed image obtained in Step 1, evaluate the safe region, and combine it with morphological filtering to obtain the binarized image of the selected safe region. Step 3: Calculate the size of the landing area. The ruggedness discriminant function value of the safe area obtained in step 2. A weighted evaluation method for landing points is constructed to evaluate all in-zone points within the safe zone, thereby achieving optimal selection of landing points within the safe zone; The specific implementation method of step 3 is as follows: Step 3.1: Using the information obtained in Step 2 Describe the flatness of the landing area; Step 3.2: Use the binarized image of the safe area obtained in Step 2. The landing area size is calculated using a spiral search algorithm based on the safety radius, resulting in the landing area size corresponding to the landing point being... ; We obtain it from the following formula: In the above formula, pixel coordinates Corresponding landing area directional safety radius length, For landing area directional safety radius length; Step 3.3, through and All in-zone points within the safe zone are evaluated, and a weighted evaluation method for landing points is constructed to optimize the landing point selection within the safe zone. The landing point evaluation function is: In the evaluation function, To evaluate the function weight coefficients, and The normalized expression for the function is represented as: in, and This represents the maximum value of the function. and This represents the minimum value of the function; The coordinates of the maximum value point calculated by equation (12) are the coordinates of the optimal landing point within the safe zone. : in, This is the set of points within the selected safe area.

2. The method for rapid safety assessment of the surface of an extraterrestrial body and selection of landing sites as described in claim 1, characterized in that: The specific implementation method of step 1 is as follows: Step 1.1: Calculate the filter kernel radius of the side window filter; The method for autonomously determining the filter kernel radius is as follows: in, The radius of the filter kernel. Represents element and The maximum value, Indicates a pair of elements Round up. Indicates a pair of elements Round down; horizontal pixel size of window and window vertical pixel size Represented as: in, This is the focal length value of the navigation camera. and For navigation camera pixel size, and Given the dimensions of the rock in the coordinate system of the optical navigation camera, The position of the rock in the coordinate system of the optical navigation camera Axial projection; the coordinate system of the optical navigation camera is defined as: Origin of the coordinate system of the optical navigation camera Located at the optical center of the navigation camera, The axis points in the direction of the navigation camera's optical axis and is perpendicular to the image plane. shaft and The plane formed by the axes is parallel to the image plane and satisfies the right-hand rule; Step 1.2: Perform side window filtering on the image acquired by the optical camera; The side-window filtering method achieves filtering by integrating the filtering results of different sub-filter kernels. The specific filtering method of the sub-filter kernels is as follows: in: This is the filtered output value of the sub-filter kernel; Here is the filtering kernel function; Let j be the pixel value of the sub-filter kernel in the image acquired by the camera, and i be the point to be filtered. Let j be the filtering weight of the sub-filter kernel with point i as the point to be filtered; Let be the sum of the filtering weights for point i to be filtered; Let S be the set of points within the sub-filter kernel; S is the set of sub-filter kernels. The kernel function parameter is defined as the angle between the sub-filter kernel and the horizontal line. The kernel function parameter is defined as the length corresponding to the sub-filter kernel; By comparing the filtered values ​​after processing by each filter kernel The original pixel value of the point i to be filtered The filtering result with the minimum Euclidean distance is obtained. The final output of the side window filtering algorithm: The result of equation (4) achieves the elimination of image noise and rocks smaller than a given size; Step 1.3: Use gradient sharpening to process the image, enhancing edge information while preserving features of flat areas; to obtain the preprocessed image.

3. The method for rapid safety assessment of the surface of an extraterrestrial body and selection of landing sites as described in claim 2, characterized in that: The specific implementation method of step 2 is as follows: Step 2.1: Use the Otsu's method to determine the image background; if the pixel is determined to be the background, then no ruggedness determination is performed; if the pixel is determined not to be the background, then proceed to step 2.

2. The optimal threshold for the Otsu's method is Represented as: in: In the above formula, The total number of pixels in the image. This represents the number of pixels corresponding to a given grayscale value. The maximum grayscale value of the image; If the pixel values ​​in the image after preprocessing in step 1 are less than Then it is the background, larger than Then it is not the background; Step 2.2: Achieve rapid assessment of regional security level using the ruggedness discrimination operator; The discriminant function of the ruggedness discrimination operator is: in: The parameters are calculated for the discrimination function of the ruggedness discrimination operator. The coordinates of the pixel to be estimated, parameters Let be the pixel coordinates surrounding the pixel to be estimated, and , , Let the set of interior points of the operator be the pixel value surrounding the pixel to be estimated. Compared with the pixel value to be estimated The difference for: therefore: In the formula, The discriminant in the discriminant function; The pixel value difference threshold is given in advance by the ground. Through the obtained Binarize the image: In the formula, The ruggedness threshold is given in advance by the ground; according to the calculation result of equation (10), the region in the preprocessed image is quickly segmented, the safe region is evaluated, and the initial binary image is obtained. ; Step 2.3: Perform morphological filtering on the initial binarized image obtained in Step 2.2 to eliminate small safe regions and spikes, resulting in a binarized image of the safe region. .

Images