149 results about "Distance transform" patented technology

The invention relates to the technical field of rice quality detection, in particular to a head rice rate online detection method and system based on machine vision, and the method comprises the steps: obtaining an initial binary image of a rice image; intelligently screening a rice grain adhesion region in the initial binarized image by using a BP neural network to obtain a to-be-segmented binarized image; segmenting each rice grain adhesion region in the to-be-segmented binary image by using a multi-distance transformation watershed segmentation method to obtain a first segmentation result image; segmenting a rice grain adhesion area in the first segmentation result image by using an angle limitation concave point matching segmentation method to obtain a second segmentation result image as a final segmentation result image; and obtaining the length of each rice grain in the rice image on the basis of the obtained segmentation result image by using a minimum bounding rectangle method so as to calculate the head rice rate of the rice image. According to the method, the segmentation accuracy is improved, the segmentation method is high in applicability, and subsequent measurement of characteristic parameters of single-grain rice is facilitated.

The invention discloses an image contour recognition method and device for a cell cluster, and relates to the technical field of biological image processing, and the method comprises the steps: converting an input cell cluster image into a binary image; carrying out expansion operation and distance transformation on the image after morphological operation, and carrying out subtraction operation on the two obtained images to obtain a third image containing an unknown region; detecting connected regions in the accurate foreground region, and marking each independent connected region, background region and unknown region to obtain a marked image; performing region segmentation on the original image to obtain segmented marked regions, generating a corresponding binary mask image, and extracting the contour of each marked region by using the binary mask image; identifying and filtering the contour of each marked area to obtain an effective contour; the size of each effective profile is converted into an equivalent sphere diameter. According to the invention, accurate segmentation of adjacent or adhesion areas in the image can be efficiently realized, and accurate counting and size measurement of cell clusters can be obtained.

The invention provides a stone gradation automatic detection method based on image processing, belongs to the technical field of image processing, and designs a two-dimensional convolution kernel capable of enhancing features according to the features that small-particle stones are bright in center, dark in edge and approximate to a circle by acquiring a gray image and a background image of a stone field through a video stream. Convolution operation is carried out on the image to improve the contrast ratio of the stone and the background, then an accurate binarized image is obtained through double-threshold segmentation and background difference processing, in order to further separate the adhered particles, the particle center is positioned by adopting distance transformation, and effective segmentation is carried out by combining a watershed algorithm. And according to the extracted particle contour, geometric parameters are calculated and the mass is estimated, so that a stone grading curve for evaluating the filling quality is automatically generated. According to the invention, rapid and non-contact automatic analysis of rockfill material grading is realized.

The invention belongs to the field of application of artificial intelligence in physical and chemical research, and provides a catalytic metal particle electron microscope image recognition method based on density estimation, which comprises the following steps: sending an electron microscope image into a backbone network for feature extraction to generate feature maps with different sampling multiplying power; inputting the feature maps with different scales into an expansion feature pyramid network, and carrying out multi-level feature fusion; sending the data to a self-attention expansion pyramid pooling module, and performing up-sampling to obtain a predicted density map; performing reverse distance transformation on the input original image to obtain a density map true value; using the predicted density map and the true value density map to calculate strong and weak loss, and weighting the two to jointly supervise the training process; performing local maximum positioning on the density map predicted by the final model through a prediction density map target positioning algorithm to obtain a target position coordinate; and finally, calculating to obtain a density and distance distribution curve of the monatomic, and marking the monatomic target in an original image. According to the invention, high-precision identification of the electron microscope image of the catalytic metal particles can be realized.

The invention discloses an edge SLAM dense map construction method based on Kirsch combined high and low thresholds. The method comprises the following steps: inputting an RGB-D picture into an improved visual odometer, extracting Kirsch edge features of a color image, performing instance segmentation on the color image of a key frame by using a BlendMASK deep neural network, and fusing semantic information into the color image. And distance transformation is constructed according to the binary image of the Kirsch edge and is used for calculating a residual error of 3D-2D registration of the key frame and the current frame. And then, carrying out pose estimation on the current frame by using 3D-2D registration of the key frame and the current frame segmented by the fusion instance. A sliding window local BA optimization algorithm based on a foreground edge is used for carrying out local optimization on the pose, and on this basis, a closed-loop detection algorithm based on a random fern descriptor is used for carrying out global pose graph optimization. And finally, based on the camera projection model and the inverse projection function, constructing a dense point cloud map. According to the method, an edge detection algorithm combining Kirsch with high and low thresholds, a BlendMASK deep neural network, a sliding window local BA optimization algorithm based on foreground edges and a closed-loop detection algorithm based on random fern descriptors are combined, so that the mapping accuracy can be improved, and a remarkable effect is achieved in the aspect of realizing an accurate dense point cloud map.

The invention belongs to the technical field of image processing, and particularly relates to an inonotus obliquus raw material granularity and uniformity detection method based on image analysis, and the method comprises the steps: obtaining an inonotus obliquus raw material gray scale image, and carrying out the preprocessing and binaryzation to obtain an adhesion body candidate object; screening particle core seed points through distance transformation in combination with an adaptive threshold strategy based on image statistical characteristics; calculating the attribution probability of a pixel to a seed point through a geodesic distance, achieving the soft segmentation of an adhesion region through a negative index model, and extracting a non-destructive segmentation line to obtain independent particles; and calculating morphological parameters of the particles, and obtaining a global uniformity index by combining the size dispersion and the shape regularity. According to the method, the accuracy of adhesion segmentation of irregular inonotus obliquus raw materials with complex textures is improved, and the robustness and accuracy of granularity detection are improved.

The invention provides a pathological image processing method and system, and the method comprises the steps: carrying out the color space decoupling of an original pathological image, separating a plurality of staining components corresponding to the optical absorption characteristics of a staining agent, carrying out the space alignment and channel superposition of the staining components, and generating a multi-channel structure diagram containing the structural characteristics of cells; inputting the multi-channel structure diagram into an adversarial network for color distribution correction to obtain a standardized red-green-blue three-primary color image; performing morphological optimization on the standardized red-green-blue three-primary color image by using an optimization process to obtain a mask, and performing Euclidean distance transformation on the mask to generate a distance map; performing channel splicing on the mask, the distance map and the standardized red-green-blue three-primary-color image, inputting a multi-task deep learning network, and outputting a probability thermodynamic diagram and a boundary probability diagram of a cell mass center; and cell segmentation is realized based on the probability thermodynamic diagram and the boundary probability diagram by utilizing a segmentation process, so that the problem of structural deformation of the image is effectively avoided.

The invention discloses a part machining surface defect visual detection method and system based on image enhancement and a binarization algorithm, and belongs to the field of machine manufacturing. The problem of how to realize high-precision and low-false-detection defect detection by optimizing an image processing flow without relying on big data training is solved, and meanwhile, the requirements of an industrial scene on real-time performance and deployment cost are met. The method comprises the following steps: inputting a part surface image and a mask image, and carrying out corrosion operation to generate a safety detection area; carrying out bottom cap transformation by adopting longitudinal and transverse morphological nucleuses, and separating scratches and noise; extracting transverse and longitudinal gradients by using a Sobel operator, and generating a multidirectional gradient feature map; performing dynamic threshold segmentation to generate a binary image, and connecting fracture scratches through closed operation; and screening longitudinal and transverse scratch contours according to the length-width ratio, carrying out distance transformation and minimum inscribed circle radius calculation, and judging the region with the radius smaller than a threshold value as a real defect. The method is mainly applied to the defect detection field.

The invention relates to the technical field of image analysis, in particular to an expressway green passing vehicle image feature extraction and rapid screening system. The expressway green passing vehicle image feature extraction and rapid screening system comprises an image acquisition unit, an image extraction unit and an image screening unit. According to the method, the image acquisition, extraction and screening processes are automated, the detection efficiency is greatly improved, the vehicle waiting time is greatly shortened, smooth passing of the expressway is guaranteed, the method based on morphological gradient and distance transformation is adopted, the complex scene that cargos are disorderly stacked and mutually shielded in a green passing vehicle container of the expressway can be dealt with, and the traffic safety of the expressway is guaranteed. In the image fusion process, the adjacent sub-regions with similar features are merged based on the feature similarity, redundant information in the image is reduced, in addition, non-green passing standard goods can be screened out, and subjectivity and uncertainty of manual judgment are avoided.

The invention provides a key point path planning method and device based on distance transformation, and the method comprises the steps: converting a grid map of a target region into a distance transformation map based on a distance definition method; a distance maximum value point is extracted from the distance transformation map to serve as a candidate key point, the candidate key point is screened to obtain a global key point, and at least two contact points exist between a circle generated with the global key point as the circle center and the distance value of the global key point as the radius and a grid representing a non-passing area; and constructing a connection relationship between the global key points, and searching and determining a navigation path based on the connection relationship between the global key points by using a # imgabs0 # algorithm. According to the method, the global key points are screened in the distance conversion map, and path search is performed based on the connection relation between the global key points, so that a more efficient robot path planning algorithm is realized.

The invention provides a robot navigation method, a storage medium, a program product and a robot. The navigation method for the robot comprises the following steps: a distance conversion step: carrying out distance conversion on a map; in the pre-planning step, the heuristic cost is calculated based on the heuristic distance, path pre-planning is carried out in the mode that the node with the small heuristic cost is selected each time, and the heuristic cost represents the shortest path from the current node to the target node plus the heuristic distance; a visual detection step: carrying out visual detection by using a line-of-sight algorithm so as to update the local father node and the remote father node; an updating step: updating the remote father node by adopting a lazy updating mode; a proceeding step: after the pre-planning is completed, proceeding from the starting node to the target node and arriving at the target node along the remote father node, and if an unknown obstacle is encountered on the way, carrying out re-planning; and an iteration step: repeating the visual detection step, the updating step and the advancing step until the target position is reached.

The invention relates to the technical field of image processing, in particular to a flower disease and insect pest monitoring method based on machine vision, and the method comprises the steps: obtaining a candidate pseudo boundary set in a gray level image of a to-be-monitored flower leaf, and obtaining a gray level distribution feature of a pixel point in a local analysis region of each candidate pseudo boundary; obtaining a vein feature index of each candidate pseudo-boundary according to the linear structure feature of each candidate pseudo-boundary and the linear structure feature of each candidate pseudo-boundary; according to the vein feature index and the direction feature of each candidate pseudo boundary, screening in the candidate pseudo boundary set to obtain at least one pseudo boundary; in the process of carrying out region segmentation on the gray level image by using the watershed algorithm, the initial distance of each pixel point after distance transformation is obtained, adaptive distance compensation is carried out on the initial distance of each pixel point according to the position relation between each pixel point and each pseudo boundary, the compensation distance is obtained, complete segmentation of a scab region is realized, and the scab image segmentation efficiency is improved. And a reliable quantitative basis is provided for flower disease and pest monitoring and prevention decision making.

The invention provides a point cloud noise reduction method based on a three-dimensional distance transformation field, and belongs to the technical field of mobile robot autonomous environment perception. According to the method, noise points are adaptively removed through dynamic resolution voxel division and a three-dimensional field intensity model, and meanwhile, a key point cloud structure is reserved. According to the method, the noisy points in the three-dimensional laser ranging data can be removed in a relatively low computing resource mode without depending on the type of a three-dimensional laser sensor and the spatial distribution characteristics and structural characteristics of the point cloud, meanwhile, the detail characteristics of the point cloud can be stored, the data accuracy is effectively improved, and the accuracy of the three-dimensional laser ranging data is improved. And the requirements of real-time positioning, scene reconstruction, subsequent map construction, scene understanding, path planning and other tasks of the mobile robot in an unknown dynamic environment on accurate three-dimensional laser data are met.

A method of performing 3D vessel tree reconstruction includes providing segmented binary angiography images, applying a distance transform to the images, and generating distance transformed binary angiography images. The set of distance transformed binary angiography images are provided to a trained 3D vessel reconstruction machine learning model capable of reconstructing 3D vessels. The 3D vessel tree reconstruction machine learning model includes a multi-stage convolutional neural network comprising a multi-stage architecture with (i) a vessel centerline stage, and (ii) a radius reconstruction stage. Resultant 3D reconstructed vessel trees may be used in performing clinical assessment of coronary vessel health, and occlusion.

The invention discloses a fry counting method based on image processing, and relates to the technical field of image processing. Comprising the steps of image acquisition, fry image target segmentation, image region division, fry target identification, preliminary fry counting, overlapped fry identification and final fry counting. Median filtering denoising is adopted, and the image quality is guaranteed; a threshold value is determined in combination with global and local maximum between-cluster variance methods, expansion and corrosion operations are assisted, fry and a background are accurately distinguished, and the processing efficiency and segmentation precision are improved through a sub-region division strategy; through connected region marking, multi-feature fusion screening and adjacent region splicing de-weighting, single fry is effectively identified, repeated counting is avoided, distance transformation and watershed algorithm accurate segmentation are adopted for overlapped fry, then a correction coefficient optimization result is introduced, and finally high efficiency and high precision of fry counting are achieved. And powerful support is provided for intelligent counting management of fishery breeding.

The invention discloses a bridging defect visual detection method for a solder paste detection instrument, a storage medium and equipment, and belongs to the technical field of surface mounting defect detection. In order to solve the problems that an existing bridging defect visual detection algorithm in a solder paste detection instrument is high in complexity and the detection accuracy needs to be improved, a solder paste area binary image is obtained based on a binary image and a solder paste area binary image, and then a solder paste printing area in the solder paste detection area binary image is removed; obtaining a binary image of the solder paste overflowing the solder paste printing area; taking a pixel of a solder paste printing area in the solder paste detection area image as a target, taking other pixels as a background, and adopting a chessboard distance to measure the length of the solder paste overflowing from the solder paste printing area; obtaining the distance between each pixel in the overflowing solder paste and the solder paste printing area through the binary image and the distance transformation image of the solder paste overflowing from the solder paste printing area, searching the maximum value and the minimum value, taking the difference between the maximum value and the minimum value as the distance of the solder paste overflowing from the solder paste printing area, and judging that the bridging defect exists if the difference exceeds a length threshold value.

A computer includes a processor and a memory, and the memory stores instructions executable by the processor to detect static environmental features in a camera image from a camera of a vehicle, generate a distance transform image of the static environmental features as detected in the camera image, and determine a pose of the vehicle based on a comparison of map data indicating the static environmental features with the distance transform image. Pixel values of respective pixels in the distance transform image indicate respective pixel distances of the respective pixels from the static environmental features in the distance transform image.

A head-removed shrimp posture detection method based on machine vision comprises the following steps: carrying out three-channel separation on a head-removed shrimp image in an HSV space, and selecting an optimal channel according to a foreground and background difference to obtain a head-removed shrimp binary image; obtaining three distance transformation heat maps of the binarized image of the head-removed shrimp, and selecting an optimal heat map according to the extraction matching degree of the central area of the shrimp head to obtain a central area image of the shrimp head; performing corner detection on the prawn head center area image, and fitting corner coordinates through a straight line to obtain a prawn head and prawn back orientation datum line; comparing the distances between the two wide sides of the minimum enclosing rectangle of the binarized image of the head-removed shrimp and the two end points of the shrimp head orientation datum line to obtain a shrimp head orientation vector; comparing the distance between the long side, tightly attached to the shrimp back, in the minimum bounding rectangle of the binarized image of the head-removed shrimp and the two end points of the shrimp back orientation datum line to obtain a shrimp back orientation vector; and the orientation vectors of the shrimp head and the shrimp back represent the attitude information of the head-removed shrimp. The method can accurately and stably detect the postures of the head-removed shrimps at low cost, and belongs to the shrimp posture image processing technology.

The invention mainly relates to the field of digital image photogrammetry. In order to solve the problems that road crack detection lacks end-to-end quantitative analysis capability and crack extraction results are fragmented, the invention provides a road crack extraction and grading method based on topological structure constraints, and the method comprises the following steps: crack end point extraction and width calculation, skeleton of a crack center line, identification of a connected region and end points, and classification of the end-to-end quantitative analysis capability of the end-to-end quantitative analysis capability of the end-to-end quantitative analysis capability of the end-to-end quantitative analysis capability. The crack width corresponding to each center point is calculated through Euclidean distance transformation; potential breakpoint pairs are screened according to attribution, distance and direction constraints, then a topological connection rule is constructed, self-adaptive repair of the fracture is achieved, and the geometric integrity of the fracture is recovered; and performing quantitative analysis on the repaired cracks, accurately calculating the length, the average width, the maximum width, the area and the geographic coordinates of each crack, and performing danger grade division on each crack according to a preset threshold value to form structured disease data and a grading result which can be directly used for road risk assessment.

The invention discloses a narrow road passing path planning method suitable for an automatic driving sweeping vehicle. The narrow road passing path planning method is particularly suitable for a complex unstructured environment in an open space. The method is based on dynamic programming mapping expansion, and comprises the following steps: firstly, constructing a grid map containing obstacles, generating a global cost graph by using breadth-first search, and constructing a Voronoi graph through obstacle distance transformation; on this basis, a heuristic path search algorithm is used to generate a preliminary path, and a potential narrow lane area in the path is extracted. And then, kinematics compensation is performed on the path by adopting an expansion mechanism based on a bicycle model, the expansion resolution is improved in a narrow lane area, and path search is guided in combination with a composite cost function. And finally, an optimal path meeting the requirements of safety, smoothness and traffic efficiency is obtained. The method improves the path planning robustness and real-time performance in a narrow channel environment, and is suitable for an automatic driving system in a complex urban environment and a semi-closed space.

The invention relates to a cellular structure segmentation method and system based on multi-round fluorescence in-situ hybridization images. The invention provides a method for automatically segmenting a cell structure in a fluorescence image. The method comprises the following steps: acquiring a fluorescence image to be analyzed as image data to be processed; training data are selected from the to-be-processed image, a cell truth value label data set is made through image labeling software, and the to-be-processed image is preprocessed; training a multi-task learning-based convolutional neural network based on the marked data set, and outputting a nearest neighbor affinity image between pixels and a probability image that the pixels belong to a boundary; on the basis of a network output result, a distance transformation watershed algorithm and a graph division algorithm are adopted to generate a cell instance segmentation result; and automatically identifying a segmentation error region in the segmentation result, artificially and selectively correcting the segmentation result, and obtaining a corrected cell boundary result through edge extraction.

The present invention relates to the technical field of railway line intelligent search, and discloses a wide-area spatial railway line GPU parallel intelligent search method, comprising the following steps: S1, establishing a comprehensive geographic information model and initializing data; S2, dividing the grid into layers; S3, parallelizing the grid scanning in the GPU to update the distance map and find a feasible route path; S4, judging whether the distance value in the distance map has changed, if changed, returning to S3, otherwise outputting the final distance map data. The present invention divides a large number of grids into layers, performs serial calculations in each thread block area, and performs parallel calculations in all thread blocks, thereby utilizing the powerful parallel processing capabilities of the GPU, and adopts a collaborative computing architecture of GPU computing and CPU control, thereby realizing the efficient application of the distance transformation method in wide-area line selection space and fine grids.

The invention relates to the technical field of intelligent detection, in particular to a method and system for detecting the cleanliness of automobile parts to be painted. The method comprises the following steps: acquiring a grayscale image of a to-be-painted automobile part; carrying out distance transformation on the preprocessed grayscale image, and calculating the saliency height of a local maximum point to screen and obtain a mark point; generating a gradient potential field map based on the gradient magnitude and the gradient confidence; generating a shape potential field map based on the distance transformation image; taking the mark point as a starting point, performing regional flooding on a fusion potential field graph obtained by weighting the gradient potential field graph and the shape potential field graph, and forming a final segmentation boundary; and finally, identifying and calculating a pollutant region based on the segmentation boundary to determine the cleanliness. According to the method, under-segmentation caused by pollutant adhesion is effectively avoided, and the accuracy of pollutant segmentation and the robustness of a detection result are remarkably improved.

The application provides a skeleton extraction method and system and a storage medium, wherein the method steps comprise: step S100, training a data set, obtaining a polygon contour and a skeleton key point of an object in an original image; taking the polygon contour as a domain constraint, triggering wave propagation connection between the skeleton key points, obtaining a preliminary skeleton, and performing an inflation operation to obtain a coarse-grained skeleton; and step S200, establishing a W-shaped network, taking a distance conversion grayscale image of the original image as an input of a first encoder, taking the coarse-grained skeleton as an input of a second encoder, taking an original real skeleton as a label, and adopting a hybrid loss function with a weight value decreasing layer by layer to complete training, so as to perform skeleton extraction. In this way, the problem of skeleton fracture caused by large shape and scale changes of different parts of the object in the process of extracting the skeleton is solved.

A method of performing 3D vessel tree reconstruction includes providing segmented binary angiography images, applying a distance transform to the images, and generating distance transformed binary angiography images. The set of distance transformed binary angiography images are provided to a trained 3D vessel reconstruction machine learning model capable of reconstructing 3D vessels. The 3D vessel tree reconstruction machine learning model includes a multi-stage convolutional neural network comprising a multi-stage architecture with (i) a vessel centerline stage, and (ii) a radius reconstruction stage. Resultant 3D reconstructed vessel trees may be used in performing clinical assessment of coronary vessel health, and occlusion.

The invention relates to the technical field of railway line intelligent search, and discloses a wide-area space railway line GPU parallel intelligent search method which comprises the following steps: S1, establishing a comprehensive geographic information model and initializing data; s2, performing hierarchical division on the grids; s3, grid scanning is carried out in parallel in the GPU to update the distance map, and a feasible line path is searched; and S4, judging whether the distance value in the distance map is changed or not, if so, returning to S3, and otherwise, outputting final distance map data. According to the method, by means of hierarchical division of massive grids, execution of serial calculation in each thread block area and parallel calculation of all the thread blocks, the powerful parallel processing capacity of the GPU is utilized, and meanwhile, a cooperative calculation framework of GPU calculation and CPU control is adopted, so that efficient application of the distance transformation method in wide-area line selection space and fine grids is achieved.

The invention relates to the technical field of image processing, in particular to a temporary loading bubble detection method and system for physical and chemical experiment evaluation, and the method comprises the steps: obtaining a color microscope image of a temporary loading, and converting the color microscope image into a smooth grayscale image; determining a gradient transformation image and a distance transformation image based on the smooth grayscale image; respectively performing size transformation on the gradient transformation image and the distance transformation image to form an enhanced image set comprising a plurality of groups of enhanced image pairs, and respectively performing logic operation on corresponding pixel points of each group of enhanced image pairs to obtain a mixed thermodynamic diagram set; performing image morphological operation, Hough circle detection and shape reduction on the mixed thermodynamic diagram set to obtain an alternative bubble edge set; deduplicating the alternative bubble edge set to obtain a bubble edge set, and filling the bubble edge set to obtain a bubble segmentation image; according to the invention, the efficiency of temporary mounting bubble detection is improved.

This invention discloses a method for calculating the tortuosity of coal-uranium reservoirs based on the maximum inscribed circle and symmetry constraints, comprising the following steps: acquiring a three-dimensional binary image of the internal structure of the coal-uranium reservoir; extracting inlet and outlet end face slices respectively, and marking the pore connectivity domains in each slice; for each pore connectivity domain, determining the center and radius of the maximum inscribed circle through Euclidean distance transformation, and filling the maximum inscribed circle in descending order of radius; setting a radius threshold, and determining the center of the filled circle that satisfies the threshold as the channel endpoint; applying the same connectivity domain marking rules, filling rules, radius threshold, and non-overlapping constraints to the inlet and outlet end faces to obtain the set of starting points and the set of ending points; searching for the optimal connectivity path from each starting point to each ending point in the three-dimensional pore space, and recording the actual length of each path; calculating the tortuosity based on the actual length of each path and the seepage length, and thus obtaining the overall tortuosity.

This invention discloses a method and system for candidate optimization, continuous repair, and quantization of crack images in complex backgrounds, belonging to the field of computer vision and digital image processing technology. To address the problems of crack recognition in complex backgrounds being susceptible to false target interference, discontinuous cracks, and difficulty in stable quantification of engineering parameters, this invention obtains the crack response probability map and determines a linked dual threshold. It retains low-threshold regions connected to the high-threshold region to generate initial candidate regions. The ratio of skeleton length to area is calculated as a thinness index for screening. Skeleton endpoints are extracted, and bridging pixel paths are generated to achieve local repair when the endpoints of different connected domains meet preset multi-condition constraints. The repaired region is then skeletonized to construct a graph structure. The number of original skeleton segments, the total number of cracks, and the total length are statistically analyzed. Width statistics and direction parameters are calculated by sampling along the skeleton using distance transformation. This effectively suppresses false targets in complex backgrounds, accurately connects discontinuous cracks, and outputs stable, accurate, and interpretable engineering parameters.