Method for realizing panoramic look-around of unmanned ship

Abstract

The invention relates to a method for realizing panoramic look-around of an unmanned ship. The method comprises the following steps: collecting a panoramic splicing image of the unmanned ship, converting the panoramic splicing image from an RGB color space to an HSV color space, obtaining a grayscale image, preprocessing the grayscale image, obtaining a superpixel region, and further generating a saliency map; weighting waves in the grayscale image according to the saliency map to generate a region of interest, and performing SLIC0 superpixel processing on the region of interest to generate a superpixel map; the method comprises the following steps of: connecting a superpixel image with an HOG (Histogram of Oriented Gradient) feature vector to obtain a composite feature vector, calculating gradient direction distribution in each superpixel region, obtaining a symmetric gradient feature, fusing the composite feature vector with the symmetric gradient feature, inputting the fused feature into a support vector machine for classification, and carrying out panorama splicing on a classification result. According to the method, the influence of insufficient illumination on the water surface and water surface ripples is fully considered, and the panoramic image splicing quality of the unmanned ship is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to a method for realizing panoramic viewing of an unmanned ship. Background Art

[0002] A ship surround view system (also known as a 360-degree surround view system or a ship panoramic monitoring system) is an advanced visual monitoring solution designed specifically for ships. The system captures images through multiple cameras installed around the ship, which are then stitched together to form a seamless panoramic view. The main purpose of this system is to provide the captain and crew with all-round visual information, especially when in narrow waters, at port or sailing, so that they can better judge the ship's position and surrounding environment.

[0003] Although ship surround view systems have made significant progress in many aspects, there are still some shortcomings and challenges in practical applications, especially in complex water environments, insufficient lighting conditions, and the effects of water ripples. The waves and currents on the water surface will cause the positions of objects in the image to change continuously, making image stitching and target recognition difficult. The fluctuations of the water surface will cause the positions of objects in the image to change continuously, making image stitching and target recognition difficult. The fluctuations of the water surface will also cause image jitter, affecting the stability and accuracy of the system.

[0004] In the existing image processing technology, the realization of the panoramic view system of unmanned ships is crucial to improving their environmental perception capabilities and the ability to perform complex tasks. However, due to the illumination changes, noise interference, water surface ripple interference and discontinuity in the image stitching process, traditional image feature extraction methods often cannot provide stable and robust feature descriptions. Summary of the invention

[0005] In order to solve the problems existing in the above-mentioned prior art, the present invention proposes a method for realizing panoramic viewing of an unmanned ship. The method fully considers the influence of insufficient lighting on the water surface, water surface ripples, etc., is more suitable for the water surface environment than the traditional method based on feature detection and feature matching, and improves the accuracy of image stitching.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] To achieve the above object, the present invention provides a method for realizing panoramic viewing of an unmanned ship, comprising:

[0008] Collecting a panoramic stitching image of the unmanned ship, converting the panoramic stitching image from the RGB color space to the HSV color space, obtaining a grayscale image, preprocessing the grayscale image, obtaining a superpixel area, and generating a saliency map according to the superpixel area;

[0009] Weighting the waves in the grayscale image according to the saliency map to generate a region of interest, performing SLIC0 superpixel processing on the region of interest to generate a superpixel map;

[0010] The superpixel image is connected with the HOG feature vector to obtain a composite feature vector, the gradient direction distribution inside each superpixel area is calculated, the symmetric gradient feature is obtained, the composite feature vector is fused with the symmetric gradient feature, the fused feature is input into a support vector machine for classification, and the classification results are stitched into a panoramic image.

[0011] Optionally, generating the saliency map includes:

[0012] Denoising and superpixel segmenting the grayscale image are performed to obtain the superpixel region, extracting feature information of the superpixel region, and performing multi-cue fusion based on the saliency value of each superpixel region calculated by combining global and local contrast to generate a preliminary saliency map; the feature information includes: color histogram, texture features and shape features;

[0013] The preliminary saliency map is smoothed, the smoothed preliminary saliency map is threshold segmented, and the background light is removed by adjusting the brightness and contrast of the background area to further optimize the boundary of the salient object to generate the saliency map.

[0014] Optionally, generating the region of interest includes:

[0015] Weighting the waves in the grayscale image according to the saliency map to obtain saliency-weighted wave coordinate features;

[0016] To set potential feature filters:

[0017]

[0018] Among them, Th hl is the maximum length of the ship, D l is the lateral distance between the left and right waves on the same ship, i.e., the length, D h is the longitudinal distance between two waves, i.e. the spacing, L1 and L2 are the coordinates of the left wave, and L'1 and L'2 are the coordinates of the right wave;

[0019] Based on the potential feature screening conditions, the potential wave pairs are screened using the lateral distance and horizontal position characteristics of the waves, and the region of interest is generated through the wave coordinate features and the wave pairs.

[0020] Optionally, generating the superpixel map includes:

[0021] Obtaining pixel points of the region of interest, using the pixel points to set a number of superpixels in the region of interest, and distributing a plurality of initial clustering centers;

[0022] Calculating the spectral distance between the initial cluster center and the pixels in the region, normalizing the spectral distance, assigning a pixel to each superpixel, and determining the maximum spectral distance from all pixels assigned to the superpixel;

[0023] The maximum spectral distance is used to update the pixel allocation of the superpixel. After the allocation is completed, the isolated pixels are merged into adjacent superpixels to generate a superpixel map.

[0024] Optionally, distributing the multiple initial cluster centers includes:

[0025] Set the initial center target neighborhood and find the pixel with the smallest gradient in the target neighborhood of each initial center as the initial cluster center.

[0026] Optionally, the formula for calculating the spectral distance between the initial cluster center and the pixels in the region is:

[0027]

[0028] Among them, D s is the comprehensive distance between the cluster center and the pixel, including the weighted combination of color distance and spatial distance, d lab is the color distance between the pixel and the cluster center in the CIELAB color space, d xy is the spatial distance between the pixel and the cluster center on the two-dimensional image plane, m is the balance factor between the color distance and the spatial distance, and S is the scale parameter of the grid size or segmentation area size of the cluster center area.

[0029] Optionally, obtaining the symmetric gradient feature includes:

[0030] The gradient direction distribution inside each superpixel region is calculated, the gradient features are divided into horizontal and vertical direction histograms, and the symmetric gradient features of the horizontal and vertical direction histograms are extracted.

[0031] Optionally, performing panoramic image stitching on the classification results includes:

[0032] A hyperplane is constructed by using a radial function, and the fused features are input as input vectors to the support vector machine for classification according to the hyperplane to obtain the classification result;

[0033] The formula of the radial function is: K(x,x') = exp(-γ||x-x'|| 2 );

[0034] where x and x' are the feature vectors of the input samples, ||x-x'|| 2 is the Euclidean distance between two samples, γ is a hyperparameter of the kernel function, controlling the width of the radial basis function;

[0035] The classification results are stitched into a panorama, and during the panorama stitching process, the overlapping areas of the images are color fused.

[0036] Optionally, the formula for obtaining the overlapping area is:

[0037]

[0038] in, is the overlapping area, W is the window function, (x i ,y j ) is the pixel coordinate, I x ,I y They are the horizontal gradient and the vertical gradient respectively.

[0039] The beneficial effects of the present invention are:

[0040] The present invention adopts a method of fusion of region-based superpixel HOG feature to extract image feature points, and uses a simple linear iterative clustering zero (SLIC0) superpixel algorithm in the preprocessing method to weight each pixel in the HOG feature generation, thereby improving the ability of adaptive feature extraction and noise reduction, and can remove the interference of ripples on the water surface on image splicing, which is more suitable for the water surface environment. The system is optimized in image feature extraction, can more significantly distinguish between texture and non-texture areas, and improves the accuracy of image splicing. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0042] Figure 1 The present invention is a flowchart of a method for realizing panoramic viewing of an unmanned ship according to an embodiment of the present invention. DETAILED DESCRIPTION

[0043] 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.

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] like Figure 1 As shown, this embodiment discloses a method for realizing panoramic surround viewing of an unmanned ship, including: collecting panoramic stitching images of the unmanned ship, converting the panoramic stitching images from the RGB color space to the HSV color space, obtaining a grayscale image, preprocessing the grayscale image, obtaining a superpixel area, and generating a saliency map according to the superpixel area; weighting the waves in the grayscale image according to the saliency map to generate a region of interest, performing SLIC0 superpixel processing on the region of interest to generate a superpixel map; connecting the superpixel map with the HOG feature vector to obtain a composite feature vector, calculating the gradient direction distribution inside each superpixel area, obtaining a symmetric gradient feature, fusing the composite feature vector with the symmetric gradient feature, inputting the fused feature into a support vector machine for classification, and performing panoramic image stitching on the classification results.

[0046] Specifically:

[0047] This embodiment discloses a method for realizing panoramic viewing of an unmanned ship, including:

[0048] The unmanned ship surround view system includes an NVIDIA development module and is connected to the camera group through a deserialization board;

[0049] The image processing module in the NVIDIA development module stitches the acquired images to form a panoramic image;

[0050] The acquired image is subjected to image distortion correction using the open source software library OpenCV, and then the method of the present invention is applied to process the image:

[0051] Step 1: Convert the rectified image from RGB color space to HSV color space, obtain a grayscale image, and extract water surface ripples by combining background illumination removal and saliency model;

[0052] In the second step, these water surface ripples are combined with a template-based method to delineate the areas containing potential features;

[0053] Step 3: Superpixel and HOG (S-HOG) feature fusion is performed in these regions;

[0054] Step 4: The vertical histogram of symmetric features of directional gradients (V-HOG) is integrated, and support vector machine (SVM) is used for classification, and the classification results are spliced ​​for the next step;

[0055] Step 5: Use grayscale fusion technology to stitch the processed images into a panoramic image.

[0056] Furthermore, generating a saliency map includes: removing noise and superpixel segmentation from a grayscale image, obtaining superpixel regions, extracting feature information from the superpixel regions, and performing multi-cue fusion based on the saliency values ​​of each superpixel region calculated by combining global and local contrast to generate a preliminary saliency map; smoothing the preliminary saliency map, performing threshold segmentation on the smoothed preliminary saliency map, removing background light by adjusting the brightness and contrast of the background region, further optimizing the boundaries of salient objects, and generating a saliency map; wherein the feature information includes: color histogram, texture features, and shape features.

[0057] The above-mentioned superpixel segmentation is equivalent to rough processing, which aims to make each superpixel contain one or more pixels and be relatively consistent in color, texture, brightness, etc.

[0058] Specifically:

[0059] The saliency-based background light removal method significantly improves computational efficiency and detection performance by leveraging a saliency model at the region level. This method not only reduces the amount of computation since the number of regions is much smaller than the number of pixels, but also extracts more informative features from the regions, such as color histograms, to better capture salient objects in the scene. In addition, by leveraging complementary prior knowledge, such as multiple cues such as color, texture, and shape, this method exhibits greater robustness and higher accuracy under complex lighting conditions. Compared with traditional patch-based methods, region-based methods can more effectively preserve the boundaries of salient objects and avoid misjudgment of high-contrast edges, thereby providing more reliable results in a variety of application scenarios.

[0060] The saliency-based background light removal method first preprocesses the input image, including noise removal and superpixel segmentation, to generate superpixel regions with similar color and texture characteristics. Next, the color histogram, texture features, and shape features are extracted from each superpixel region, and the saliency value of each region is calculated by combining global and local contrast. Through multi-cue fusion, a preliminary saliency map is generated, and it is smoothed to remove noise and discontinuous areas. Then, the saliency map is divided into foreground and background using a threshold segmentation method (such as the Otsu method), and the influence of background light is removed by adjusting the brightness and contrast of the background area. Finally, the boundaries of salient objects are further optimized through edge detection and post-processing steps such as morphological operations to generate the final saliency map, ensuring that the boundaries of salient objects are clear and the background light has been effectively removed. This series of steps ensures the efficiency and accuracy of the method under complex lighting conditions.

[0061] Further, generating the region of interest includes: weighting the waves in the grayscale image according to the saliency map to obtain saliency-weighted wave coordinate features;

[0062] To set potential feature filters:

[0063]

[0064] Among them, Th hl is the maximum length of the ship, D l is the lateral distance between the left and right waves on the same ship, i.e., the length, D h is the longitudinal distance between two waves, i.e. the spacing, L1 and L2 are the coordinates of the left wave, and L'1 and L'2 are the coordinates of the right wave;

[0065] Based on the potential feature screening conditions, the lateral distance and horizontal position characteristics of the waves are used to screen potential wave pairs, and the region of interest is generated through the wave coordinate characteristics and wave pairs.

[0066] Specifically:

[0067] The template-based method first preprocesses the captured image to reduce noise and enhance wave features. Then, the waves in the image are weighted using the saliency map to generate saliency-weighted wave coordinate features. With these features, regions of interest (ROIs) are generated to screen potential wave pairs using the lateral distance and horizontal position characteristics of the waves. Specifically, the lateral distance D of two waves on the same ship is l Must be less than the maximum value of the ship Th l , and these two waves are basically at the same horizontal position. That is:

[0068] Although this method may lead to some false detections, it can effectively avoid missed detections and provide enough candidate regions for the subsequent verification stage. In the verification stage, other features (such as wave shape, texture, and motion characteristics) are further used to fine-tune the candidate regions and finally determine the accurate wave position. This series of steps ensures the robustness and accuracy of the method in complex water surface environments.

[0069] Furthermore, generating a superpixel map includes: obtaining pixel points of an area of ​​interest, setting a number of superpixels in the area of ​​interest using the pixel points, and distributing a plurality of initial cluster centers; calculating the spectral distance between the initial cluster center and the pixels in the area, normalizing the spectral distance, assigning pixels to each superpixel, and determining the maximum spectral distance from all pixels assigned to the superpixel; using the maximum spectral distance, updating the pixel assignment of the superpixel, and after the assignment is completed, merging the isolated pixels into adjacent superpixels to generate a superpixel map.

[0070] Furthermore, distributing a plurality of initial cluster centers includes: setting an initial center target neighborhood, and searching for a pixel with the smallest gradient in the target neighborhood of each initial center as the initial cluster center.

[0071] Specifically:

[0072] The S-HOG (Superpixel-based Histogram of Oriented Gradients) method significantly improves the spatial consistency, adaptability, discrimination, and robustness of features by fusing a region-based superpixel approach into HOG features. Specifically, the method uses the Simple Linear Iterative Clustering Zero (SLIC0) superpixel algorithm to weight each pixel in HOG feature generation without the need to input a compactness factor, thereby automatically calculating superpixels. This method aims to improve spatial consistency, handle changes in ship orientation, provide adaptive feature extraction, reduce noise, enhance discrimination, exploit contextual information, and create a more comprehensive and robust feature representation for accurate feature recognition under diverse and challenging conditions.

[0073] The S-HOG method uses the SLIC0 algorithm to process the image and then connects it with the HOG directional gradient histogram method for feature extraction, and finally fuses the two to obtain a super-pixel level feature representation. Different from the traditional method of directly applying HOG on the entire image, the present invention extracts HOG features at the super-pixel level generated by SLIC0, which means that each super-pixel has an independent HOG feature descriptor, which can better capture local structural information while reducing the amount of calculation.

[0074] Feature fusion: Combine the superpixel boundary information generated by SLIC0 with the HOG feature to form a richer feature representation. Add statistical information such as the color mean and variance of the superpixel on the basis of the HOG feature, or use the shape and size of the superpixel to enhance the distinguishability of the feature.

[0075] In the process of implementing the S-HOG method, the SLIC0 algorithm is first used to perform superpixel segmentation on the input image. This process adaptively adjusts the content of superpixels to optimize the performance of texture and non-texture areas while maintaining low computation and memory consumption. Subsequently, the images of the two channels are resized to a standard size of 32×32 pixels, and the size of each HOG cell is set to 8×8 pixels, and each block contains 4 such cells:

[0076] SLIC0 superpixel processing process:

[0077] 1. Convert image color space RGB to HSV;

[0078] 2. Calculate the initial grid spacing. Assuming that the image has N pixels, the expected number of superpixels generated is K, and the expected side length of each superpixel is

[0079] 3. Cluster center initialization:

[0080] K initial cluster centers are evenly distributed in the image. To prevent these centers from falling on the edge of the image or noise points, SLIC0 searches for the pixel with the smallest gradient in the 3x3 neighborhood around each initial center as the final cluster center.

[0081] Use the Sobel operator or other edge detection methods to calculate the gradient of the image to help select the most appropriate initial clustering center. The gradient calculation only needs to consider the information of the L channel, that is, the brightness information of the image, to simplify the calculation.

[0082] 4. Distance calculation:

[0083] Each pixel is represented as a five-dimensional vector [l, a, b, x, y], where l, a, b are components in the HSV color space, and x, y are the position coordinates of the pixel. For a given distance between the cluster center and the pixel, the following formula is used for calculation:

[0084]

[0085] Among them, D s is the comprehensive distance between the cluster center and the pixel, including the weighted combination of color distance and spatial distance, d lab is the color distance between the pixel and the cluster center in the CIELAB color space, d xy is the spatial distance between the pixel and the cluster center on the two-dimensional image plane, m is the balance factor between the color distance and the spatial distance, and S is the scale parameter of the grid size or segmentation area size of the cluster center area.

[0086]

[0087] Among them, (y p -y c ) is the image coordinate of the pixel, (x p -x c ) are the image coordinates of the cluster center.

[0088] Unlike standard SLIC, SLIC0 introduces normalization of spectral distances. After each iteration, the maximum spectral distance is determined from all pixels assigned to this superpixel. In the next iteration, the pixel assignments to the superpixel are updated, and when evaluating the potential assignment of a pixel to a superpixel, the spectral distance from the pixel to the superpixel is divided by the maximum spectral distance of the current superpixel. This helps to compare the spectral differences between different superpixels more fairly.

[0089] 5. Processing isolated pixels. After the iteration, there may be some isolated pixels that do not belong to any large superpixel segment. In order to ensure the connectivity of each superpixel, SLIC0 will process these isolated pixels and merge them into the adjacent largest superpixel. Specifically, the algorithm scans the image from top to bottom and from left to right, tries to search for unlabeled pixels, and adds the pixels in its four neighborhoods that belong to the same old superpixel to the search queue to form a new connected superpixel.

[0090] 6. Generate superpixel map. SLIC0 outputs an image segmentation result containing multiple compact and regular superpixels. The color and texture inside each superpixel are relatively consistent, and the boundary fits the object outline in the original image as closely as possible.

[0091] The 324-dimensional HOG feature vector is constructed by evenly dividing the gradient direction of the pixel into 9 histogram channels. Finally, the 324-dimensional HOG feature vectors generated by the grayscale image and the red image are merged to form a 648-dimensional composite feature vector. In order to further improve the classification performance of the model, a support vector machine (SVM) based on the radial basis function (RBF) kernel is used to learn and construct the optimal decision surface, thereby achieving effective verification and classification of ripples.

[0092] This superpixel segmentation aims to further simplify and optimize the superpixel segmentation process and generate high-quality segmented pixels. The SLIC0 algorithm is introduced, which has the following advantages: automatic parameter adjustment, emphasis on generating high-quality superpixels, higher computational efficiency, and better robustness.

[0093] Furthermore, obtaining the symmetric gradient features includes: calculating the gradient direction distribution inside each superpixel region, dividing the gradient features into horizontal and vertical direction histograms, and extracting the symmetric gradient features of the horizontal and vertical direction histograms.

[0094] Further, performing panoramic image stitching on the classification results includes: constructing a hyperplane using a radial function, inputting the fused features as input vectors to a support vector machine for classification according to the hyperplane, and obtaining classification results;

[0095] The formula of the radial function is: K(x,x') = exp(-γ||x-x'|| 2 );

[0096] where x and x' are the feature vectors of the input samples, ||x-x'|| 2 is the Euclidean distance between two samples, γ is a hyperparameter of the kernel function, controlling the width of the radial basis function;

[0097] The classification results are stitched into a panorama, and during the panorama stitching process, the overlapping areas of the images are color fused.

[0098] Specifically:

[0099] Connect the average feature vector of the superpixel region with the HOG feature vector to generate a composite feature vector F S-HOG =[F S ,F HOG ], F S is the superpixel feature vector, F HOG is the Hog eigenvector.

[0100] V-HOG feature fusion:

[0101] Calculate the gradient direction distribution inside each superpixel area and divide the gradient features into horizontal and vertical histograms. Extract the symmetric gradient feature F in the vertical direction V-HOG .

[0102] F S-HOG With F V-HOG Further integration: F final =[F S-HOG ,F V-HOG ], and reduce the dimensionality through t-SNE dimensionality reduction technique.

[0103] SVM Classification:

[0104] It is necessary to pre-train an SVM classifier model, use the radial function (RBF) to construct a hyperplane, and transform the composite feature F corresponding to each superpixel area S-HOG As the input vector, the category label feature points and non-feature points are output, and the radial function is as follows: K(x,x') = exp(-γ||x-x'|| 2 );

[0105] Among them, x and x' are the feature vectors of the input samples, ||x-x'|| 2 is the Euclidean distance between two samples, and γ is a hyperparameter of the kernel function, which controls the width of the radial basis function. Usually γ>0, the larger its value, the narrower the radial function and the more sensitive it is to local data.

[0106] There may be two-dimensional differences between the images to be spliced, resulting in two-dimensional discontinuity in the splicing position of the image, which will affect the visual effect. Usually, the overlapping area of ​​the image is color-fused to solve this problem. The present invention uses a linear interpolation fade-in and fade-out method to fuse the spliced ​​images:

[0107] Assume that the overlapping interval of images I1 and I2 on the x-axis is [x min ,x max ], the present invention first matches the feature points of the classification results using the formula:

[0108]

[0109] in, is the overlapping area, W is the window function, (x i ,y j ) is the pixel coordinate, I x ,I y They are the horizontal gradient and the vertical gradient respectively.

[0110] Assume λ1,λ2 are two eigenvalues ​​of the matrix, and the basis for interest point detection is: in, This criterion has a strong response to points of interest, and k is generally taken as 0.04~0.06.

[0111] Calculate the color value I0 of the overlapping area:

[0112] I0(x,y)=βI1(x,y)+(1-β)I2(x,y);

[0113] Where β is the gradient factor, and its value is β=(x min -x)(x max -x), is the trace of the matrix is the determinant of the matrix, x min ,x max are the starting and ending positions of the region on the x-axis respectively.

[0114] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for realizing panoramic viewing of an unmanned ship, characterized in that: include: Collecting a panoramic stitching image of the unmanned ship, converting the panoramic stitching image from the RGB color space to the HSV color space, obtaining a grayscale image, preprocessing the grayscale image, obtaining a superpixel area, and generating a saliency map according to the superpixel area; Weighting the waves in the grayscale image according to the saliency map to generate a region of interest, performing SLIC0 superpixel processing on the region of interest to generate a superpixel map; The superpixel image is connected with the HOG feature vector to obtain a composite feature vector, the gradient direction distribution inside each superpixel area is calculated, the symmetric gradient feature is obtained, the composite feature vector is fused with the symmetric gradient feature, the fused feature is input into a support vector machine for classification, and the classification results are stitched into a panoramic image.

2. The method for realizing panoramic viewing of an unmanned ship according to claim 1, characterized in that: Generating the saliency map comprises: Denoising and superpixel segmenting the grayscale image are performed to obtain the superpixel region, extracting feature information of the superpixel region, and performing multi-cue fusion based on the saliency value of each superpixel region calculated by combining global and local contrast to generate a preliminary saliency map; the feature information includes: color histogram, texture features and shape features; The preliminary saliency map is smoothed, the smoothed preliminary saliency map is threshold segmented, and the background light is removed by adjusting the brightness and contrast of the background area to further optimize the boundary of the salient object to generate the saliency map.

3. The method for realizing panoramic viewing of an unmanned ship according to claim 1, characterized in that: Generating the region of interest comprises: Weighting the waves in the grayscale image according to the saliency map to obtain saliency-weighted wave coordinate features; To set potential feature filters: Among them, Th hl is the maximum length of the ship, D l is the lateral distance between the left and right waves on the same ship, i.e., the length, D h is the longitudinal distance between two waves, i.e. the spacing, L1 and L2 are the coordinates of the left wave, and L'1 and L'2 are the coordinates of the right wave; Based on the potential feature screening conditions, the potential wave pairs are screened using the lateral distance and horizontal position characteristics of the waves, and the region of interest is generated through the wave coordinate features and the wave pairs.

4. The method for realizing panoramic viewing of an unmanned ship according to claim 1, characterized in that: Generating the superpixel map includes: Obtaining pixel points of the region of interest, using the pixel points to set a number of superpixels in the region of interest, and distributing a plurality of initial clustering centers; Calculating the spectral distance between the initial cluster center and the pixels in the region, normalizing the spectral distance, assigning a pixel to each superpixel, and determining the maximum spectral distance from all pixels assigned to the superpixel; The maximum spectral distance is used to update the pixel allocation of the superpixel. After the allocation is completed, the isolated pixels are merged into adjacent superpixels to generate a superpixel map.

5. The method for realizing panoramic viewing of an unmanned ship according to claim 4, characterized in that: Distributing the multiple initial cluster centers includes: Set the initial center target neighborhood and find the pixel with the smallest gradient in the target neighborhood of each initial center as the initial cluster center.

6. The method for realizing panoramic viewing of an unmanned ship according to claim 4, characterized in that: The formula for calculating the spectral distance between the initial cluster center and the pixels in the region is: Among them, D s is the comprehensive distance between the cluster center and the pixel, including the weighted combination of color distance and spatial distance, d lab is the color distance between the pixel and the cluster center in the CIELAB color space, d xy is the spatial distance between the pixel and the cluster center on the two-dimensional image plane, m is the balance factor between the color distance and the spatial distance, and S is the scale parameter of the grid size or segmentation area size of the area where the cluster center is located.

7. The method for realizing panoramic viewing of an unmanned ship according to claim 1, characterized in that: Acquiring the symmetric gradient feature includes: The gradient direction distribution inside each superpixel region is calculated, the gradient features are divided into horizontal and vertical direction histograms, and the symmetric gradient features of the horizontal and vertical direction histograms are extracted.

8. The method for realizing panoramic viewing of an unmanned ship according to claim 1, characterized in that: Panoramic image stitching of classification results includes: A hyperplane is constructed by using a radial function, and the fused features are input as input vectors to the support vector machine for classification according to the hyperplane to obtain the classification result; The formula of the radial function is: K(x,x') = exp(-γ||x-x'|| 2 ); where x and x' are the feature vectors of the input samples, ||x-x'|| 2 is the Euclidean distance between two samples, γ is a hyperparameter of the kernel function, controlling the width of the radial basis function; The classification results are stitched into a panorama, and during the panorama stitching process, the overlapping areas of the images are color fused.

9. The method for realizing panoramic viewing of an unmanned ship according to claim 8, characterized in that: The formula for obtaining the overlapping area is: in, is the overlapping area, W is the window function, (x i ,y j ) is the pixel coordinate, I x ,I y They are the horizontal gradient and the vertical gradient respectively.

Images