55 results about "Edge operator" patented technology

The present invention discloses a method (400) and apparatus for AVC intra prediction for encoding digital video comprising multiple images. The method includes the following steps: generating (410) edge direction information for each intra-frame coding block of the digital image; selecting (420) the most likely intra-frame prediction mode according to the generated edge direction information for rate-distortion optimization. The above-mentioned edge direction information can be generated by applying at least one edge operator to the digital image. The edge direction histogram can be used to accumulate the magnitudes of pixels in 15 directions within the coding block. The method may further comprise the step of intra-coding (430) a block of the digital image with the selected most likely intra-frame prediction mode.

A method of eliminating unwanted steps at edges in image or graphic representations in the line raster, in particular in on-line operation having the steps:a) Application of an edge operator to an image portion for coarsely ascertaining at least one rastered edge configuration,b) Determining the position of at least a first pixel from the amount of those pixels which form the rastered edge configuration or adjoin said rastered edge configuration,c) Approximation of a straight line for ascertaining a probable configuration of the unrastered image edge in the proximity of the first pixel,d) Ascertaining a criterion from the approximation straight line and the position of the first pixel for mixing a color X to the color C in the first pixel considered, ande) Mixing the ascertained color X to the color C in the first pixel considered.

The invention discloses a moving object segmentation method based on optical flow field clustering. The method is characterized in that optical flow fields of an image sequence are clustered to effectively detect and segment single or multiple moving object(s) in a complex image background. An object area is segmented by utilizing moving internal epipolar constraint and C-mean value cluster algorithm to obtain a segmentation image; a detailed target area is obtained from the segmentation image by utilizing a Canny edge operator, and an edge image is obtained; and the segmentation image is merged with the edge image according to the flow velocity in the optical flow field, and complete single or multiple moving objects is / are detected. Thus, the moving object can be reliably segmented and detected under the condition that a camera moves.

The invention discloses a method for intelligently measuring distances of moving vehicles in front of vehicles. The method comprises the following steps of: S1, acquiring a front road video image, andcarrying out preprocessing such as image clipping, image graying, image filtration and image binarization on the image; S2, strengthening road image edge information by adoption of a Canny edge operator; S3, detecting a lane line by adoption of a Hough transformation method; S4, segmenting a vehicle bottom shadow, establishing a rectangular frame of a possible vehicle, namely, a region of interest (RIO), and carrying out accurate detection on the vehicle; S5, establishing a visual projection model, and calculating a pixel coordinate value of a midpoint on the bottom edge of the image and a pixel coordinate value of a midpoint at the bottom edge of the RIO; S6, establishing a demarcation template to demarcate parameters in a camera; S7, establishing a geometric distance measurement model;and S8, obtaining road plane coordinate system coordinates of the midpoint at the bottom edge of the image and the midpoint at the bottom edge of the RIO, and combining the demarcation result of parameters in the camera to calculate a distance between the vehicle and the front vehicle according to a distance calculation formula.

The invention discloses a drone thermal infrared image-based coal fire identification method. The method mainly comprises the following steps that: a thermal infrared camera is mounted on a rotor drone pan-tilt holder, and a drone completes the acquisition work of the thermal infrared images of a mine area according to a designed flight route; the acquired thermal infrared images are preprocessed;the thermal infrared camera is calibrated through a laboratory black body radiator, the DN values of the images are converted into apparent radiance values; all the thermal infrared images are subjected to aerial triangulation adjustment orienteering, and a thermal infrared orthoimage is obtained through mosaic, and the surface temperature of the mining area is obtained through inversion on the basis of an atmospheric transport equation and the Planck function; and a high gradient graph is generated according to the Sobel edge operator, the high gradient graph is refined into skeleton lines,a high temperature area and the high gradient lines are superimposed, the temperature average value of the overlapping position of the high temperature area and the high gradient lines is adopted as the fire area segmentation threshold value of the mining area, and the threshold value is adopted to identify the coal fire area of the mining area. The coal fire identification accuracy of the methodcan be as high as 96.7%. With the method of the invention adopted, the rapid and accurate identification and drawing of coal fires under the conditions of complex mining areas can be realized at relatively low cost.

The invention discloses a wood counting method based on contour optimization. The method comprises the steps that dimension conversion is carried out on a wood image; a wood target image is extracted from the conversed image through a color model, and edge information in the image is extracted through an edge operator; the extracted wood target image and the edge information are merged to acquire a pre-segmented image; the outline of a candidate target is extracted from the pre-segmented image, and extracted outer contour and inner contour are respectively corroded and expanded for optimization; and the above process is repeated until the number of extracted outer contours is invariant and the inner contour is not present. According to the optimization process, a false target and the adherence of targets are eliminated for the wood candidate target of the pre-segmented image; and the extracted wood target can be directly used for counting. According to the invention, the validity of the method is experimentally verified.

The invention discloses an intelligent traffic light based on the signal control processing technology and a signal control method. The intelligent traffic light comprises a video detection device, an image processing module, an optimizing timing calculation module, a control module and a traffic signal display module, wherein the video detection device, the image processing module, the optimizing timing calculation module, the control module and the traffic signal display module are connected in sequence. Firstly, the video detection device shoots practical road conditions, vehicle queuing images are obtained, boundary operators are extracted through the image processing module according to the digital image processing technology, and a current vehicle queue length is worked out according to a built vehicle queue length model; then, vehicle queue length information is transmitted to a communication port of a single chip microcomputer through wireless communication, and the optimal green light duration is worked out according to the current vehicle queue length through a linear algorithm embedded in the single chip microcomputer. According to the intelligent traffic light based on the signal control processing technology and the signal control method, the traffic light changes intelligently according to traffic flow, and therefore the purposes of reducing traffic pressure, saving energy and reducing emission are achieved.

The invention relates to an in-situ observation method for gas arc electrode sputtering ablation, which comprises the following steps of: irradiating the surface of an electrode by adopting high-powerlaser, and filtering arc light by utilizing a narrow-band filter to obtain an image for representing an electrode sputtering ablation behavior in an arc combustion process; using a Gaussian-Laplacianedge operator to filter the original image to achieve complete separation of arc light and an electrode, and obtaining the edge of the splashing liquid drop; Reconstructing the splashing liquid dropand extracting position and shape information of the splashing liquid drop based on an image gray threshold binaryzation method, and obtaining a motion track of the splashing liquid drop through framesuperposition; And obtaining the spatter ablation rate based on a multi-target particle matching algorithm. Through the combination of an optical imaging means and a digital image processing technology, visualization and in-situ measurement of the electrode ablation behavior in the arcing process are achieved, testing of the gas arc ablation behavior is expanded, the method provides a direct anduseful scientific basis for in-depth and complete understanding of the gas arc-electrode mechanism and the process of sputter ablation..

The invention discloses a wireless sensor network abnormal event detection method and system based on secondary mixed compression. The method comprises the steps that S1. primary compression is performed on original data sequences by a compression sensing method; and secondary compression is performed on the compressed sequences by a piecewise linear fitting method so that state edge operators are obtained; S2. edge amplitude and edge intensity of each data point in the compressed sequences are calculated, and new edge point sequences are formed by selecting edge points of low interpolation error from edge point sequences; S3. characteristic value sequences are obtained according to edge point sequences; S4. local accessible density and local abnormal factors are calculated; and S5. event detection is performed in corresponding time sequence intervals according to the size of the local abnormal factors after mixed compression. Search efficiency and abnormal data accurate positioning capacity of wireless sensor network nodes can be enhanced so that abnormal events can be more efficiently and rapidly found; and timeliness of abnormity detection can be enhanced so that energy and communication bandwidth can be greatly saved.

The invention discloses a PCB secondary and multiple accurate punching method based on visual tracking. After the image information of the PCB is obtained through the vision system, the image information is subjected to binaryzation through a cvThreshold function; calculating a communication domain which exists on the processed image by utilizing a FindBlobs function, and filtering the communication domain which is smaller than the area; accurately positioning the edge by means of a Zernike moment edge operator in the possible edge points after all possible edge points are quickly found through a Sobel operator, and calculating to obtain the center position coordinates of the Mark points or the prefabricated hole according to the edge coordinates. The data volume of the points needing to be calculated is greatly reduced, and the speed of determining the center coordinates is increased, and the accuracy is not lost; by adopting a multi-time punching mode, the center coordinates of the prefabricated hole are positioned again during each punching, and the deviation problem possibly caused by single positioning is corrected, the drill bit is a grinding process of the edge of the prefabricated hole punched previously during multiple times of punching, and the precision is far higher than that of the single punching.

The invention relates to a target celestial body image culmination circle judgment method for deep space detection autonomous navigation. The method comprises: carrying out edge extraction on a target celestial body image, wherein a traditional edge operator method is adopted to carry out edge point extraction; selecting a candidate arc segment; carrying out fitting judgment; and carrying out culmination circle supplement judgment. A fitting precision of a center position of a target celestial body image can be directly affected if a culmination circle of a space target is not judged and is eliminated. With the present invention, a problem of target celestial body culmination circle judgment in interplanetary detection can be solved, and culmination circle judgment is adopted to find a true edge point to carry out target celestial body center position fitting so as to enhance a positioning precision of the celestial body image target center; and important engineering use values are provided for developments of high robustness, high refresh rate and high precision navigation celestial body sensors.

The invention relates to a fine-grained classification method for fashionable women's wear images based on component detection and visual characteristics, belonging to the field of computer vision andimage application. The invention firstly detects the parts of the body part of the input fashion women's wear image to be classified and the image in training. Secondly, the detected images of fashion women 's wear are extracted respectively, and the HOG, LBP, color histogram and edge operator of fashion women' s wear images are trained to get the feature extracted images. Then, the defined visual feature descriptors are matched with the extracted four underlying features, and multi-class SVM is used to supervise the learning and training of the fine-grained classifier model. Finally, throughthe fine-grained classifier after training, the fashion women 's clothing image undergone the features classifying achieves fine-grained classification, fashion women' s clothing image classificationresults are output. The detection and classification method adopted by the invention has high accuracy.

The present invention discloses an infrared target detection method based on an improved Tri-edge operator, which has low complexity and high detection accuracy. The method comprises the following steps: (10) background prediction: predicting to obtain a background prediction image according to a to-be-detected original image; (20) residual image extraction: subtracting the background prediction image from the original image to obtain a residual image; (30) increase of the contrast of the image: after adding the residual image to the original image, doubling the grayscale value to obtain a high-contrast image; (40) image noise suppression: carrying out Wiener filtering to obtain a noise-suppressed image; (50) enhanced image acquisition: superimposing the noise-suppressed image with the original image to obtain an enhanced image; (60) edge image extraction: using the edge extraction operator to extract edge information of the noise-suppressed image; (70) image fusion: fusing the enhanced image and the edge image; and (80) target identification: according to the adaptive threshold, identifying the target to obtain an infrared detection target.

The invention discloses an automatic focusing device. The device comprises a lens, a lens drive component, a first control module and a second control module. The first control module is applicable for obtaining a first frequency response result of pixels of an original image in a first mode and outputting a noise reduction image based on the first frequency response result of a passband part. The second control module is applicable for extracting a space domain edge operator at an interested image position of the noise reduction image, correspondingly multiplying the space domain edge operator and the pixel values at the interested image position according to edge directions and summing, thus obtaining a space domain result. The first control module is also applicable for obtaining a second frequency response result of interested pixels of the original image in a second mode. The lens drive component is applicable for taking the lens position when an expected space domain result / the second frequency response result is realized, as a focusing position, thus driving the lens. According to the device, under the conditions of low contrast textures and weak light, the focusing accuracy and the focusing speed of a photographed object are improved.

The invention discloses a full automatic extraction technology of high spatial resolution cultivated land based on depth learning. The technology of the present invention is guided by the edge extracted by the traditional Canny edge operator and is based on the depth learning theory, trains the HED depth learning model through a large number of edge sample label data, and improves the network layer number, pool size and other parameters of the model. And then the new network model is used to extract the cultivated land image, and finally, the boundary extracted by Canny edge operator is refined and eliminated to achieve the extraction of the skeleton edge of cultivated land. Compared with the traditional manual land extraction, the technology of the invention can effectively improve the production efficiency of the land extraction, and furthermore can ensure the unification of the land edge precision.

Improved triangle management in triangular meshes uses a data structure having two fields to store data for each triangle in the triangular mesh. The first field is a set of three vertices for the triangle and the second field is a set of three edges, each edge corresponding to one of the three vertices. Each of the three edges is an identification of a next or subsequent edge that is encountered when performing a traversal (e.g., in a counterclockwise direction) about the corresponding vertex. According to one aspect, three operators are defined to assist in management of the triangular mesh. These operators are a make edge operator, a splice operator, and a swap operator, and are selectively invoked to both add triangles to the triangular mesh and remove triangles from the triangular mesh.

The invention discloses an image denoising method based on an adaptive weight total variational model. An adaptive weighted total variational model is constructed. A non-local mean filter is used to process the noisy image to obtain an estimated de-noisy image. After preprocessing, four edge operators are used to detect the noisy image so as to obtain and edge detection image, and the adaptive normal parameters are defined according to the four edge operators. The pixels of the edge detection image are reset according to the adaptive normal form parameters to obtain the weight matrix of the noisy image. The denoised image is obtained by substituting the estimated denoised image, the adaptive normal form parameters and the weight matrix into the adaptive weighted total variational model. The invention can self-adaptively separate the edge region and the smooth region of the image according to the image characteristics, and the image texture and the edge are kept while noise is removed,thereby avoiding the generation of the step effect.

Methods, apparatuses, and computer program products are herein provided for enabling hand gesture recognition using an example infrared (IR) enabled mobile terminal. One example method may include determining a hand region in at least one captured frame using an adaptive omnidirectional edge operator (AOEO). The method may further include determining a threshold for hand region extraction using a recursive binarization scheme. The method may also include determining a hand location using the determined threshold for the extracted hand region in the at least one captured frame. The method may also include determining a fingertip location based on the determined hand location. Similar and related example apparatuses and example computer program products are also provided.

The invention discloses a generation method of a computer oil painting and a system thereof. The generation method comprises the following steps: receiving a source image to be processed, determining whether the source image is a color image or a grey-scale image, and if the source image is the color image, converting the color image into the grey-scale image; taking a sampling point in a grid mode on the grey-scale image, and carrying out random offset on the sampling point; using an edge operator in a horizontal direction and an edge operator in a vertical direction to calculate a gradient at the sampling point; calculating a stroke direction; using an edge operator in a 45 DEG direction and an edge operator in a 135 DEG direction to calculate edge intensity in a window; determining a stroke radius; taking a pixel value of a corresponding position of the sampling point on the source image as a pixel value of a present brush; drafting the sampling point according to the stroke direction, the stroke radius and the pixel value of the present brush. By enforcing a technical scheme of the invention, when using a computer to generate the oil painting, the method is simple, and efficiency of realization is high.

The invention provides an image detection method and device. The image detection method comprises obtaining an image to be detected, carrying out edge processing on the image to be detected in dependence on an edge detection algorithm, and obtaining a first image; and carrying out directional correction on the first image in order to obtain a target image. With respect to the image detection method and device, an edge operator with a certain width is employed to replace a conventional edge operator, isolated noise can be eliminated, and a real edge can be gathered; furthermore, a plurality of points to be selected corresponding to the edge are divided into groups in dependence on the distance and / or direction, and therefore the points to be selected which are distributed scattered show some regularity, in dependence on the shown regularity, a fitting line corresponding to the edge can be fitted, the quality of the candidate fitting line is improved, and furthermore the image edge positioning accuracy is improved. Furthermore, the image is subjected to directional correction in a planar and three-dimensional manner by means of a multi-edge operator and space grid correction, the correction precision is improved, and the calculated amount is reduced.

The invention relates to an automatic focusing device. The automatic focusing device comprises a lens, a lens driving part, a noise reduction module and a focusing control module; on the basis of the lens, an original image can be generated through light guide of a shot object; the noise reduction module is suitable for obtaining a frequency response result of a pixel point of the original image and outputting a noise reduction image on the basis of the frequency response result of a passband part; the focusing control part is suitable for extracting a spatial domain edge operator in an interested image position of the noise reduction image, and correspondingly multiplying and summarizing the spatial domain edge operator and a pixel value in the interested image position according to the edge direction, such that a spatial domain result is obtained; and the lens driving part is suitable for driving the lens so as to obtain the expected spatial domain result, and driving the lens by taking the position of the lens when the expected spatial domain result is achieved as the focusing position. By means of the invention, the focusing accuracy and the focusing speed of the shot object can be increased under conditions of low contrast texture and weak light.

The invention discloses an obstacle extraction method based on camera data of a train AEB system. The obstacle extraction method comprises the following steps: S1, acquiring an RGB color image in realtime; s2, performing graying, denoising and enhancement processing on the RGB color image; s3, carrying out image track positioning and obstacle crossing detection; s4, carrying out the primary detection of a central obstacle; s5, establishing a detection window; s6, carrying out secondary detection on the middle obstacle; and S7, calibrating the actual distance of the obstacle. An image sequenceon a track in front of a train advancing route is obtained through a monocular camera mounted in front of the train. A simple search algorithm and image morphology processing are used for positioninga rail and detecting a cross-over obstacle. A Canny edge operator extraction mode is used to carry out initial detection on an obstacle arranged in an orbit. A detection window is established, a graylevel co-occurrence matrix method is adopted to carry out secondary detection on a centrally-arranged obstacle, and finally, an IPM model is utilized to calibrate the distance between a train and theobstacle, so that data extraction of the obstacle is realized, and finally, automatic emergency braking of the train is realized.

The invention discloses a curve identification and car height adjusting method for an electronic-controlled air suspension (ECAS) system. The curve identification and car height adjusting method comprises the steps that real-time road information of a curve, where a car is about to go into, in front of the car is obtained through a camera, images are subjected to morphological filtering and region-of-interest dividing treatment, and road traffic signs and lane line information are extracted; the road traffic signs are processed through a method based on RGB color space processing and Canny edge operators, grayscale conversion, threshold segmentation and a Bezier curve model are used for three times of fitting to process a lane line, curve model data are obtained, and model data accuracy isimproved through dynamic detection; and finally, the safety car height for curve passing is calculated through the curve model data to serve as the target height, a height adjusting signal is outputto the ECAS system through the car height adjusting method, and thus the car is adjusted to the safety car height before curve passing. Compared with the prior art, the curve identification and car height adjusting method has the characteristics of high identification accuracy and scene pertinence, and travelling safety during curve passing of the car is ensured.

Improved triangle management in triangular meshes uses a data structure having two fields to store data for each triangle in the triangular mesh. The first field is a set of three vertices for the triangle and the second field is a set of three edges, each edge corresponding to one of the three vertices. Each of the three edges is an identification of a next or subsequent edge that is encountered when performing a traversal (e.g., in a counterclockwise direction) about the corresponding vertex. According to one aspect, three operators are defined to assist in management of the triangular mesh. These operators are a make edge operator, a splice operator, and a swap operator, and are selectively invoked to both add triangles to the triangular mesh and remove triangles from the triangular mesh.

The invention discloses a multi-function edging device. One end of an adjusting arm is movably riveted to a base and can be fixed via a stop bolt; the other end of the adjusting arm is connected with a protective baffle; the protective baffle can be arranged between the eyes of an edging operator and an edging part for shielding by rotationally adjusting the adjusting arm; a sliding groove is formed in the side wall of the base in the circumferential direction; an arc splash plate is close to and fitted to the side wall of the base; the bottom end of the arc splash plate is embedded in the sliding groove, so that the arc splash plate can slide in the circumferential direction of the base; and a magnifying lens extending outwards is also arranged on the adjusting arm, is connected with the adjusting arm through a gooseneck, and can be placed between the protective baffle and a millstone by adjusting the gooseneck. The multi-function edging device is simple in structure and convenient to use, and can effectively prevent edging chippings from splashing all around or even enter the eyes of the edging operator, so that the safety of the edging operator and surrounding materials is guaranteed; and moreover, waste materials can be recycled, so that the resource utilization rate is improved.

Methods, apparatuses, and computer program products are herein provided for enabling hand gesture recognition using an example infrared (IR) enabled mobile terminal. One example method may include determining a hand region in at least one captured frame using an adaptive omnidirectional edge operator (AOEO). The method may further include determining a threshold for hand region extraction using a recursive binarization scheme. The method may also include determining a hand location using the determined threshold for the extracted hand region in the at least one captured frame. The method may also include determining a fingertip location based on the determined hand location. Similar and related example apparatuses and example computer program products are also provided.

A surface defect detection system for hollow blown containers is disclosed. According to the system, a machine vision technology and image processing algorithm are combined, a Gaussian filter is utilized for removing disturbance, an image with clear edge features is acquired through Sobel operator, a profile curve is fitted by the utilization of the least square method, and programming is completed through LabVIEW and a detection result is obtained. The image algorithm is stable, and the system has a good detection effect on most defects.

The invention discloses a method for intelligently detecting moving vehicles in front of vehicles. The method comprises the following steps of: S1, acquiring a front road video image, and carrying outpreprocessing such as image clipping, image graying, image filtration and image binarization on the image; S2, strengthening road image edge information by adoption of a Canny edge operator; S3, detecting a lane line by adoption of an improved Hough transformation method; S4, segmenting a vehicle bottom shadow, and carrying out morphological correction, wherein only a series of shadow horizontalline areas remain in the corrected image; S5, establishing rectangular frames of possible vehicles, namely, regions of interest (RIO), on shadow horizontal lines; and S6, fusing vehicle texture features and vehicle symmetric features to carry out assumption verification on front vehicles, so as to realize accurate detection of vehicles.

The invention relates to a feature fusion-based multi-module unsupervised learning retinal vessel segmentation system. The system comprises an image denoising and enhancing module, a feature extraction and fusion module, a multi-module learning module and a synthesis and result analysis module, wherein the image denoising and enhancing module is used for denoising color fundus images and enhancingcontrasts of the color fundus images; the feature extraction and fusion module is used for extracting invariant moment features, Hessian moment features, Gabor wavelet features, phase equalization features and Candy edge operator features of pixels of the color fundus images, and fusing the features into feature vectors; the multi-module learning module is used for segmenting the feature vectorsof the pixels of the color fundus images into a plurality of modules and respectively clustering the modules; and the synthesis and result analysis module is used for synthesizing and comparing clustering results. The system disclosed by the invention is capable of covering the shortages that the training samples are difficult to obtain and the training time is long as supervised retinal vessel segmentation methods require experts to manually mark vessels.