37 results about "Contour set" patented technology

The application discloses a method and a device for recognizing text areas in an image. A specific embodiment of the method comprises: acquiring a color value and position information of each pixelin the image to be recognized; clustering the pixels based on the color value of each pixel, wherein the color values of the pixels in each pixel category are the same or similar; for each pixel category after clustering, determining the contour of each connected area constituted by the pixels in the pixel category to obtain a contour set; and merging the contours based on the color value and position information of each contour in the contour set to obtain respective text areas in the image. The embodiment improves the accuracy of text line sequence recognition in image character recognition.

The invention discloses a three-dimensional model search method based on mesh segmentation. The method comprises the following step of analyzing and constructing a segmentation field through a hierarchical spectrum, and particularly comprises the steps of judgment of concave vertexes, constructing of a Laplacian matrix, matrix decomposition, selection of low-frequency feature vectors, generation of sub feature vectors, weight calculation of the sub feature vectors, and constructing of an edge symbol matrix. Contour lines are sampled in the segmentation field and are grouped and merged through a grouping-merging algorithm to obtain a plurality of candidate contour sets, the final segmentation boundary is determined according to the weight of each contour line in the candidate contour sets, and three-dimensional models are automatically segmented. Three-dimensional model mixing feature description sub-matrixes are obtained by calculating the feature description sub-matrix of each segmentation block of the three-dimensional models, the similarity of the mixing feature description sub-matrixes of each three-dimensional model in a three-dimensional model database to be searched and a target three-dimensional model to be searched is calculated, the similarity values of the three-dimensional models are ranked and output from low to high, and the three-dimensional model searching is achieved.

The invention discloses a method for identifying pushing action based on a two-dimensional planar camera and a system. The method includes acquiring video images containing persons, and dividing hand detecting areas; realizing binarization processing for divided hands, and acquiring corresponding contours of the hands; comparing the contours of a divided current hand with a standard hand contour set in advance, and judging whether push action of the hand is performed; determining that the pushing action of the hand is performed when contour similarity is larger than a set threshold value T, extracting the hand area of the hand performing the pushing action, extracting the center of the minimum rectangle surrounding the contour of the hand, and creating a historical movement graph; and identifying the pushing action of the hand for the built historical movement graph by the aid of orientation gradient vectors. The method is simple in algorithm and high in action identification rate, and simultaneously burden on users in terms of memory is reduced.

A method, apparatus, article of manufacture, and computer readable storage medium provide the ability to update a fill region of a computer drawing. A topological recipe representing the fill region is defined. The recipe includes a region set (of regions to be filled where the regions are areas bound by contours), a contour set (of contours that are a chain of curve segments), and curve segments. A final fill area of the fill region based on the topological recipe is computed. The final fill area is displayed and the curve segments are then modified. In response to the modification of the curve segments, the final fill area is dynamically updated and displayed.

A method and a system for localizing at least one circular object in a video frame captured by an image capture unit are disclosed. The method includes the steps of: A. performing edge detection to obtain a plurality of image edge points in the video frame; B. extracting a contour set of image contours by linking neighboring edge points; C. generating a plurality of circle samples from the extracted contour set by a structural sampling process; and D. localizing the at least one circular object from the generated circle samples by calculating a respective fitness score for each of the circle samples to measure the supporting evidence of the image edge points on a neighborhood of each of the circle samples, and by choosing the circle sample(s) whose fitness score(s) is / are greater than a threshold value as the circular object(s) found in the video frame.

The invention discloses a photovoltaic module glass crack method, device and apparatus and a storage medium. The module includes: binarizing an image of a photovoltaic module to obtain a binarized image; extracting a contour of the binarized image to obtain an initial contour set C; filtering all initial contours according to the area of the outer rectangle and the number of corner points to obtain the contour set G of the photovoltaic module; traversing other initial contours except the photovoltaic component contours in the initial contour set C, obtaining an initial contour contained in anyof the photovoltaic component contours, and generating a candidate contour set S; detecting corner points of all candidate contours, and calculating an included angle between an adjacent corner pointvector of each candidate contour and an edge of a photovoltaic component contour to which each candidate contour belongs; when the number of included angles meeting a predetermined angle condition issmaller than a preset threshold ratio, judging that cracks exist in the glass of the corresponding photovoltaic module. Through the invention, glass crack detection of the photovoltaic module can berealized automatically, accurately and timely.

Provided is a number identification method based on a SVM. The method comprises the following steps of: 1) acquiring a number plate image I1; 2) performing color preprocessing on the image I1 to obtain a grayscale image I4; 3) performing Canny operator edge detection on the image I4 to obtain an edge image I5; 4) performing digital area rectangle framing on the edge image I5 to obtain a contour set k<3> including numbers, recognizing the contour set k<3> including numbers by using a trained number classifier in combination with the original image I1 to obtain specific numbers. The method may substitute for conventional manual classification, increases marathon photograph classification speed, is good adaptive to the flexible material of a number plate and high in recognition rate.

The invention relates to a method and a device for screen location in visual inspection. The method comprises the steps of obtaining a complete machine RGB image of a to-be-inspected apparatus; converting the complete machine RGB image into an HSV image; extracting the V domain image information in the HSV image and performing binarization processing on the V domain image information to obtain a corresponding binarization image; processing the obtained binarization image by using a preset contour extracting function to obtain a contour set; acquiring the contour with the maximum area in the contour set as the screen contour of the to-be-inspected apparatus and locating the screen of the to-be-inspected apparatus according to the screen contour. The method and the device can realize online inspection, requires no line stop or departure from a line, thereby improving the inspection efficiency of visual inspection of apparatuses.

An emergency area confinement and safety system includes a central control unit for integrating all the other units in the system. A detector unit for detecting accidents occurred in a building unit. An accident determination unit receives signals from the detectors and determining whether an accident has occurred. The accident determination unit has an accident determination logic unit for determining an accident being occurred; a contour setting unit for defining a contour as an unsafe area. The contour setting unit receives the signals from the accident determination unit to know which section has accident and then determines an unsafe area according to the detecting cells; and a guiding unit for guiding people to run away from the accident area to a safety area; and a display unit in the central control unit for displaying the states.

The invention provides an object surface sampling point set boundary characteristic identification method based on a local sample projective contour shape, and belongs to the field of digital design and manufacturing. The method is characterized in that each sample point of a sampling point set is subjected to normal estimation, all sample points are added with a label at the same time; through the labels, a sample point can be judged whether the sample point belongs to a boundary sample point state or a non-boundary sample point state, the labels of all sample points are initially set to be in a boundary sample point state; for each sample point of the sampling point set, if the labels of the sample points are in a boundary sample state, a local sample projective contour is established for the sample points along the normal direction of the sample points; then the projective points of the sample points along the normal direction are verified to judge whether the projective points fall into the contour set composed of recessed points and protruded points of local sample projection, if the projective points of the sample points do not fall into the set, the labels of the sample points are changed into a non-boundary sample point state, and finally the sub-set composed of sample points labeled as a boundary sample point state is output. The provided method can rapidly recognize the boundary sample points of an object surface sampling point set.

The invention provides an image-based gastrointestinal three-dimensional reconstruction and visualization method, which comprises the following steps of: processing an image acquired by a camera on anactive capsule machine to obtain a gastrointestinal intermittent wrinkle contour set; fitting and connecting the intermittent edges belonging to the same fold in the contour set to generate a complete two-dimensional contour, and constructing a reverse mapping model by utilizing camera internal parameters to convert two-dimensional contour information into a three-dimensional space, so as to obtain a three-dimensional local model of the gastrointestinal structure; and splicing the local models to realize global visualization of the gastrointestinal structure. The method can quickly and accurately reconstruct the gastrointestinal structure.

The invention discloses a contour line-based three-dimensional entity generation method and device. The method comprises the following steps: acquiring a contour line set for generating a target three-dimensional entity, processing the contour line set to obtain sample points of each contour line, combining the sample points according to a combination mode to obtain a sample point combination, determining a target sample point combination based on screening parameters of each sample point combination, performing enhanced point insertion on the target sample point combination to obtain enhanced points corresponding to the target sample points, and finally generating a three-dimensional entity image according to all the sample points and all the enhanced points. According to the method, sample point combinations which will generate sharp edges and corners in the process of generating the three-dimensional entity by the contour lines can be accurately found out, and enhancement points are inserted for the sample point combinations so as to avoid the sharp edges and corners.

The invention discloses a visual identification method for a water pump impeller ring. The method comprises the following steps: S1, collecting an original image of a water pump impeller ring horizontal plane by using an industrial camera; S2, converting the collected original image into a grayscale image; S3, performing binarization and morphological processing on the grayscale image to obtain avisual image; S4, performing contour set retrieval on the visual image, and traversing all contours; S5, creating a reference image with the same size as the original image and a gray value of 0, drawing a bounding circle of each contour in the step S4 on the reference image, and drawing each contour to obtain accurate contour information; S6, calculating the angle relation of the circle center ofeach contour surrounding circle obtained by the reference image based on the circle center of the contour surrounding circle obtained by the original image, and converting the pixel coordinates of the circle center into a robot coordinate system. By means of the method, the water pump impeller ring can be accurately and rapidly recognized, human factors are avoided, efficiency is improved, and subsequent grabbing of a robot is facilitated.

The invention provides a stroke separation method based on a Chinese character contour, which comprises the following steps: extracting the contour of a Chinese character, determining an inner contourand an outer contour, and obtaining a combined contour set; extracting contour candidate feature points of stroke separation; aiming at any candidate feature point; calculating the distance between any two feature points in the feature point set, the input vector and the output vector of each feature point, and initializing a feature point pair list; calculating a cosine included angle corresponding to the two feature points if the distance between the first feature point and the second feature point is not greater than a threshold value between any two different feature points; if a preset condition is met, adding the first feature point and the second feature point into a feature point pair set; and performing stroke separation based on the contour set, the feature point set and the feature point pair set. By applying the embodiment of the invention, through feature point position detection and angle detection, the accuracy of effective feature point screening is improved, and the accuracy of independent stroke extraction is improved.

The invention belongs to the technical field of digital image processing, and discloses a method, device and equipment for positioning characters in cartoon bubbles, and a readable storage medium. Themethod comprises the steps: obtaining a digital cartoon image, sequentially carrying out the gray scale and binarization processing, and carrying out the contour detection of a binary image to obtaina contour set; extracting a bubble set and a character set from the interior of the contour set according to the relationship hierarchy of the contours; removing interference contours in the character set according to the contour area, then collecting the circumscribed rectangle set of each remaining contour in the character set, removing the interference circumscribed rectangle in the initial character rectangular set according to the background color around the circumscribed rectangle, and establishing the mapping relation between the circumscribed rectangles in the character rectangular set and the contours in the bubble set; and expanding each circumscribed rectangle in the text rectangular set, and storing the position information of the expanded circumscribed rectangles. Position information can be provided for character backfilling, the backfilling effect can be improved by optimizing the character position, and better reading experience is provided for a cartoon reader.

The invention discloses an automobile hub identifying and matching method comprising the following steps: creating a contour pixel data database; calculating the average height M0 of the contour pixeldata in the database; calculating the square root mean square error MSE of the arithmetic mean sum of the difference squares of the Y coordinates of the pixels of the contour and the average height M0; and obtaining contour data of a specific position of a to-be-classified hub on an assembly line, matching the contour data with the contour data of hubs in the contour pixel data database one by one, and outputting the hub model with the minimum matching error as a final result. The automobile hub identifying and matching method is a non-contact real-time online measuring method based on a linear laser sensor. A contour is pre-matched according to the statistical features of the contour, and a hub is finally matched in a pre-matched contour set. There is no need to match all contours. The matching efficiency is improved. The method can replace manual measurement, and can be used to quickly and accurately identify different hub models.

A method, apparatus, article of manufacture, and computer readable storage medium provide the ability to update a fill region of a computer drawing. A topological recipe representing the fill region is defined. The recipe includes a region set (of regions to be filled where the regions are areas bound by contours), a contour set (of contours that are a chain of curve segments), and curve segments. A final fill area of the fill region based on the topological recipe is computed. The final fill area is displayed and the curve segments are then modified. In response to the modification of the curve segments, the final fill area is dynamically updated and displayed.

The invention discloses an IHC digital preview image identification and organization foreground segmentation method and system, and the method comprises the steps: extracting a binary mask image M1 comprising an organization part and an impurity part for an inputted IHC digital preview image; dividing all connected domains in the binary mask map M1 into a large-area connected domain contour set qband a small-area connected domain contour set qs; and for each connected domain contour in the large-area connected domain contour set qb, corresponding the connected domain contour to a corresponding region in an original IHC digital preview, extracting features of the corresponding region, classifying the features through a machine learning model to remove impurity connected domain contours, and leaving connected domain contours of tissues. According to the method, the tissue foreground can be segmented for the IHC digital preview which is slightly dyed and has the foreground tissue color almost close to the slide background, and the method has the advantages of being high in recognition accuracy and good in recognition effect.

The invention discloses a method and a relevant device to eliminate image contour noise points, which are beneficial to solve the problem in the prior art that a currently available financial device is often interfered by noises when it processes images. The method comprises: acquiring the contour points of an image; obtaining a contour point set A; removing some discrete value from the set A; obtaining a contour set B; and using the contour points in set B for Hoff straight line fitting to determine the image contour.

The invention discloses a video stream processing method and device. The method comprises the following steps: determining a target contour of a preset rigid body in an image frame in a video stream, and processing an image of the preset rigid body in the target contour, wherein the step of determining the target contour of the preset rigid body in the image frame in the video stream comprises the steps of determining a first upper vertex, a first lower vertex, a first left vertex and a first right vertex of the preset rigid body in the image frame, and matching a predetermined reference feature value set of the preset rigid body according to the first upper vertex, the first lower vertex, the first left vertex and the first right vertex, and determining a target contour of the preset rigid body in the image frame according to the matching result and a predetermined reference contour set of the preset rigid body. According to the invention, the contour of the preset rigid body in the video stream can be accurately determined in real time.

The invention discloses a Chinese character geometric structure feature migration method based on a composite contour skeleton, and the method comprises the steps: obtaining the geometric structure relation of a corresponding composite contour skeleton feature point set between fonts, constructing the mapping relation between composite contour sets of the same target word of different font files, and constructing the mapping relation between a single composite contour set and the skeleton feature points; forming the composite contour of the variant glyph of the variant glyph font to be generated, writting the corresponding coding position of the font file after format conversion. The problem that a user can only utilize the existing variant font library when using the text information hiding system is solved.

The invention discloses a three-dimensional model search method based on mesh segmentation. The method comprises the following step of analyzing and constructing a segmentation field through a hierarchical spectrum, and particularly comprises the steps of judgment of concave vertexes, constructing of a Laplacian matrix, matrix decomposition, selection of low-frequency feature vectors, generation of sub feature vectors, weight calculation of the sub feature vectors, and constructing of an edge symbol matrix. Contour lines are sampled in the segmentation field and are grouped and merged through a grouping-merging algorithm to obtain a plurality of candidate contour sets, the final segmentation boundary is determined according to the weight of each contour line in the candidate contour sets, and three-dimensional models are automatically segmented. Three-dimensional model mixing feature description sub-matrixes are obtained by calculating the feature description sub-matrix of each segmentation block of the three-dimensional models, the similarity of the mixing feature description sub-matrixes of each three-dimensional model in a three-dimensional model database to be searched and a target three-dimensional model to be searched is calculated, the similarity values of the three-dimensional models are ranked and output from low to high, and the three-dimensional model searching is achieved.

The invention provides a filling path generation method based on a skeleton line, which comprises the following steps: step 1, biasing an input two-dimensional contour set according to a set path spacing to obtain a biased contour; step 2, generating a connected region set by using the offset contour set; step 3, obtaining a skeleton line of each connected region; step 4, segmenting a single connected region according to the skeleton line to obtain a plurality of sub-regions; step 5, generating sub-filling paths in each sub-region according to the set path spacing; and step 6, connecting the sub-filling paths to obtain a filling path of the whole connected region. According to the method, the problem that more empty paths exist in a conventional common path filling method can be avoided, and a solution is provided for path planning with special process requirements. The method is suitable for areas of various shapes, has the advantages of being simple in algorithm and good in universality, and has a certain promotion effect on development and popularization of the filling path in multiple fields.

The invention discloses an object contour extraction method. The method comprises the steps of acquiring the absolute temperature of an object to form a thermodynamic diagram; linearly constraining the thermodynamic diagram temperature to a first preset real number range to form a thermodynamic diagram real number matrix; converting the thermodynamic real number graph matrix into a grey-scale graph matrix; converting the grey-scale map matrix into a binarized grey-scale map matrix in a second preset real number range; filtering noise points of the binarized grey-scale map matrix; converting the binary grayscale image matrix into a binary image in a first preset real number range; extracting all contours in the binary image, and simultaneously generating a contour hierarchy table which is used for representing an inclusion relation between the image contours; eliminating all inner contours through the contour hierarchy table to acquire a frame of measurement peripheral contour set. Theinvention further discloses an object contour extraction system, an object contour prediction method and an object contour prediction system. According to the invention, all contours in the blurred thermal image can be accurately extracted or predicted, and the contour information acquisition precision in the blurred thermal image is improved.

The invention discloses a method for identifying pushing action based on a two-dimensional planar camera and a system. The method includes acquiring video images containing persons, and dividing hand detecting areas; realizing binarization processing for divided hands, and acquiring corresponding contours of the hands; comparing the contours of a divided current hand with a standard hand contour set in advance, and judging whether push action of the hand is performed; determining that the pushing action of the hand is performed when contour similarity is larger than a set threshold value T, extracting the hand area of the hand performing the pushing action, extracting the center of the minimum rectangle surrounding the contour of the hand, and creating a historical movement graph; and identifying the pushing action of the hand for the built historical movement graph by the aid of orientation gradient vectors. The method is simple in algorithm and high in action identification rate, and simultaneously burden on users in terms of memory is reduced.

The invention relates to a method for quickly generating a building floor 3D model in real time. The method comprises the following steps: extracting triangular data in a facade model, and constructing a BVH acceleration structure; acquiring an external facade contour line segment set Se from the BVH acceleration structure; processing the external facade contour line segment set Se, extracting floor contours and nested room contours, and storing the floor contours and the nested room contours into a contour set Sp; detecting a nesting condition between contours in the contour set Sp, and extracting contour data P of an indoor room; and triangularizing the contour data P to obtain a triangular serial number array used for drawing a floor 3D model, and generating a building grid model based on the triangular serial number array. According to the method, the building floor model can be quickly generated in real time under the condition that the external facade model exists, the workload of modeling personnel is greatly reduced, real-time operation and quick feedback to users can be realized, and detail adjustment can be performed in time. The invention further relates to a device for quickly generating the building floor 3D model in real time, a storage medium and equipment.

The invention discloses a method and system for IHC digital preview image recognition and tissue foreground segmentation. The invention includes extracting a binary mask image M1 including tissue parts and impurity parts from an input IHC digital preview image; All connected domains in Figure M1 are divided into large-area connected domain contour sets q b and the set of small-area connected domain contours q s ;For large-area connected domain contour collection q b For each connected domain outline in , it is corresponding to the corresponding area in the original IHC digital preview image, the features of the corresponding area are extracted and the features are classified by the machine learning model to remove the connected domain outline of impurities and leave the connected domain outline of the organization . The invention can segment the foreground of the tissue from the IHC digital preview image with very light staining and the color of the foreground tissue is almost close to the background of the glass slide, and has the advantages of high recognition accuracy and good recognition effect.

The invention provides a method for identifying the boundary feature of a sampling point set on a physical surface based on the contour shape of a local sample projection, which belongs to the field of digital design and manufacturing, and is characterized in that: normal estimation is performed on each sample point in the sampling point set and all Add a mark to the sample point, through which it can be distinguished whether the sample point is a boundary sample point state or a non-boundary sample point state, and the marks of all sample points are initially set to the boundary sample point state; for each sample point in the sampling point set, if its If it is marked as a boundary sample point, then construct a local sample projection contour along the normal direction of the sample point, and verify whether the projection point along the normal direction of the sample point falls into the contour set formed by the concave and convex points of the local sample projection, if If it does not fall into the set, the mark of the sample point is modified to be a non-boundary point state, and the subset formed by the sample points marked as the state of the boundary sample point is output. The invention can quickly identify the boundary sample points of the object surface sampling point set.

The invention provides a pipeline three-dimensional model construction method and system based on standard attitude angle correction, and the method comprises the steps of setting a standard collection attitude angle of a pipeline robot, and recording the standard collection attitude angle as a standard attitude angle; setting the pipeline robot carrying the posture determining equipment and the obtaining equipment in the pipeline; enabling the pipeline robot to obtain a collection attitude angle of the pipeline robot through attitude determination equipment at a set position; comparing the acquired attitude angle of the pipeline robot with a standard attitude angle, and acquiring the cross section contour of the pipeline at the current position after directly acquiring or adjusting the acquired angle according to a comparison result; and establishing a pipeline three-dimensional model based on parallel contours through the pipeline cross section contour set of each position. According to the invention, the pipeline three-dimensional model with a unified acquisition view angle can be established, the influence of distortion and errors generated in the acquisition process on the accuracy of the pipeline three-dimensional model is avoided, and a good model basis is provided for pipeline internal detection.