114 results about "Sobel edge detection" patented technology

The invention discloses a driver fatigue detecting system based on a smart mobile phone. A video colleting module (1) collects images of the face and the eyes of a driver. An image preprocessing module (2) removes image noise through the gray scale adjustment of the image pixel points. A face detecting and locating module (3) achieves face detection and locating based on image pixel nonlinear color transformation. A face following module (4) improves the speed of detection of the face area of the driver through the correlation between every two adjacent images. An eye detecting module (5) conducts binarization on the images based on the improved horizontal Sobel edge detection method. An eye feature parameter extracting module (6) extracts the pupil opening degree feature parameters. A fatigue judging module (7) judges whether the driver is fatigue when driving a vehicle based on the PERCLOS method. According to the driver fatigue detecting system based on the smart mobile phone, the face detection is rapid, efficient and accurate, the requirement for the resolution is low, a camera of an ordinary mobile phone can achieve image collection and detection, popularization and application are convenient, cost is low, and the accuracy is high.

The invention provides a method for evaluating beautifulness of human faces based on video. The method is achieved by following steps: (1) capturing the face information by video; (2) effectively detecting the face according to the captured information by a cascade classifier based on haar-like characteristics; locating the eyes and the lips on the basis of the face detection, including detecting eyes by adopting a Sobel edge detection operator and a region growing method and determining the boundary of the lips according to the projection of red pixel dots in the face image; (4) extracting the face contour by adopting Snake model; and (5) comparing the extracted information with a face beautifulness standard library to determine the face beautifulness evaluation result. The method can achieve the real-time face beautifulness evaluation by capturing the face video information by a computer, and training to obtain the beautifulness classification standard according to a large amount of face samples.

The invention provides a dense population estimation method and a system based on multi-feature fusion. The method comprises the following steps: partitioning an image into N equal sub-blocks; performing hierarchical background modeling on the image by using a method based on a CSLBP (Center-Symmetric Local Binary Pattern) histogram texture model and mixture Gaussian background modeling, extracting the foreground area of each sub-block subjected to perspective correction, detecting the edge density of each sub-block in combination with an improved Sobel edge detection operator, and extracting four important texture feature vectors in different directions for describing image texture features in combination with CSLBP transform and a gray-level co-occurrence matrix; performing dimension reduction processing on the extracted population foreground partition feature vectors and texture feature vectors through main component analysis; inputting the dimension-reduced feature vectors into an input layer of a nerve network model, and acquiring the population estimation of each sub-block through an output layer; adding to obtain the total population. The dense population estimation method and system have high accuracy and high robustness, and a good effect is achieved in the population counting experiment of subway station monitoring videos.

The invention discloses a video object segmentation method based on time domain fixed-interval memory compensation, which comprises the steps of: firstly, detecting an initial motion change domain by using a symmetrical frame frame-difference accumulation method and a four-order moment of a frame-difference image; secondly, compensating the detected initial motion change domain by using a time domain fixed interval memory compensation, and further integrating to form a global motion memory mother board; in a spatial domain, more accurately detecting by using a Sobel edge detection operator to obtain all edges in a current frame; thirdly, carrying out the temporal-spatial fusion, thereby extracting a complete and fine motion object outline; and finally, filling to obtain a video motion object template. The invention relates to a novel parallel spatial fusion automatic segmentation method which effectively solves the problems that serious loss inside a video object usually occurs in the spatial fusion and blocking (covering / exposing) can not be avoided when the motion domain is detected by using frame difference, and greatly improves the accuracy, the universality and the speed.

The invention discloses a fake plate detection method based on license plate identification and vehicle feature matching. The method comprises the following steps that: extracting a monitoring equipment frame image, and carrying out graying on a source image; adopting Sobel edge detection to position a license plate; adopting a morphological processing image to enable regions to be communicated soas to bring convenience for extracting the outline of the license plate; setting an aspect ratio to accurately extract areas; through hough transformation and vertical projection, carrying out license plate correction and character segmentation; using a neural network to identify segmented characters to obtain license plate information; migrating an AlexNet neural network frame, and carrying outclassification through the identification of the depth feature of the color; and applying a KNN (K-Nearest Neighbor) algorithm to be combined with database system information to detect a fake plate situation. By use of the method, vehicle identification accuracy is guaranteed, a high-accuracy convolutional neural network is directly migrated to serve as a basic framework, cost and expenditure aresmall, the method can be quickly realized on a computer platform, cost is small for arranging a license plate identification and vehicle identification system on a large scale, and feasibility is high.

The invention discloses an image characteristic registration based geometrical fine correction method for an aviation multispectral remote sensing image, comprising following steps of: 1) utilizing a Sobel edge detection operator to carry out edge extraction on both an aviation multispectral remote sensing image to be registered and a normal incidence high-resolution satellite image which is taken as a standard; 2) utilizing a Harris angular point detection method to detect angular points of the aviation multispectral remote sensing image to be registered and the normal incidence high-resolution satellite image which is taken as the standard; 3) carrying out rough correlation between the two images through a correlation method; 4) carrying out fine correlation between the angular points of the two images through calculating supporting strength; 5) carrying out back calculation to obtain a multinomial coefficient according to a multinomial model; and 6) carrying out gray level re-sampling by adopting a bilinear interpolation to obtain a registered image. According to the image characteristic registration based geometrical fine correction method for the aviation multispectral remote sensing image, disclosed by the invention, the geometrical correction problem of the aviation multispectral remote sensing image lacking of a ground reference point can be better solved and the geometrical correction precision of an aviation multispectral scanner is improved, so that an aviation remote sensing technology can be better applied to production livings of national economy.

The invention discloses an image-processing-based cable sheath thickness measuring method, which overcomes the defects of low precision, high error, low efficiency and the like of the traditional geometrical measurement method. The method comprises the following steps of: pre-processing an acquired cable sheath slice image by using a K-nearest neighbor filter, processing the image by adopting an improved Sobel edge detection method, thinning edge pixel points, extracting inner and outer contours of the cable sheath, performing sub-pixel positioning by adopting a polynomial interpolation method, and finally detecting the inner and outer edge thicknesses of the cable sheath of national standard specifications. The method for measuring the cable sheath has the advantages of high measurement precision, low measurement error, large dynamic range, good flexibility and the like.

The invention provides a device and a method for automatically measuring driving sight distance before road turning. The method comprises the following steps: firstly, performing gray equalization, filtering and Sobel edge detection on an acquired image sequentially through an industrial personal computer, determining an interested area of the detected image, and performing threshold segmentation on the interested area by an optimal threshold value method to obtain a target area and a background area; secondly, acquiring a lane line equation of the target area; thirdly, extracting a calculation feature point of the driving sight distance on the lane line according to the lane line equation; finally, according to the pixel coordinate values of the calculation feature point of the driving sight distance, performing inverse perspective protection conversion to acquire the distance between the calculation feature point of the driving sight distance under the world coordinate system and the vehicle. The measuring method is simple in operation, convenient in detection, low in use cost, high in measuring precision and intuitive in calculation result, and manual measurement by a laser range finder, a measuring tape and the like is not needed.

The invention discloses a dynamic image moving object detection method. The method includes firstly, reading two images of time-consecutive frame and respectively subjecting the images to Gaussian filtering; secondly, calculating consecutive frame differential images to obtain a binary image with moving information; thirdly, utilizing the improved Chan-Vese algorithm to detect the moving object of the image; and finally, utilizing the improved region growing algorithm to detect the accuracy of the moving object. The two consecutive frame differential images are used as a seed pixel point set of the moving object, the Sobel edge detection operator is used for detecting the pixels around the seed pixel, the pixels around the seed pixel are marked as moving object pixels if the pixels around the seed pixel are detected as edge pixels, the detected edge pixels and the moving object detection result are mixed, and accordingly the moving object can be detected accurately.

The invention discloses a Sobel edge detection and image block brightness feature-based blind image tampering forensic method. The method is characterized by comprising the following steps of: converting a to-be-detected suspicious image into a grayscale image; carrying out convolution processing on the grayscale image I (i, j); obtaining a gradient image G (i, j) of the image; carrying out threshold value segmentation on the G (i, j) to obtain the gradient image G (i, j); carrying out binary processing on the gradient image G (i, j) to obtain a binary image W (i, j); carrying 1 pixelation on the binary image; and judging the similarity of two sub-image sets. According to the method, blocking processing is carried out on image sets; through brightness mean value sorting, the forensic algorithm efficiency can be effectively improved; and through comparing the similar brightness values of image blocks, the correctness is further improved and the image forensic efficiency is further improved. According to the method, the problem that the tampered images cannot be correctly detected due to cloning and tampering behaviors of large-scale zooming is solved; and through normalized image block brightness values, the detection result of cloned images with different brightness values is further improved.

The invention relates to a multi-feature deep learning-based vehicle detection method and apparatus. The method comprises the steps of collecting a standard captured vehicle image by an intelligent network camera; extracting feature information of three channels for the vehicle image, wherein the feature information mainly includes grayscale information, Sobel edge detection in an X direction and Sobel edge detection in a Y direction; inputting the feature information of the three channels to a combination feature convolutional neural network model obtained by pre-training to perform vehicle detection; and outputting a detection result. According to the method and the apparatus, for appearance information and shape information of a vehicle, the three feature channels are extracted, original RGB color features are abandoned, priori knowledge of people to the features is added, and a combination feature convolutional neural network method is used, so that the calculation amount required for training can be reduced, the detection time can be shortened, and the vehicle locating accuracy is improved.

An image is focused onto an image sensor array by efficient movement of an adjustable focus lens by a lens focus motor. A minimum of three focus positions is required to determine an estimated best focus position of the lens focus motor. After which a sweep of the focus positions surrounding the estimated best focus position is performed to verify and determine the best actual focus position of the lens focus motor. A sharpness value of the image being focused comprising a Sobel edge detection or other detections means provide sharpness data to evaluate image sharpness at various lens focus motor positions.

The invention discloses an underwater target detecting image enhancement method with contrast limited adaptive histogram equalization. The method comprises the following steps of calculating a four-directional Sobel edge detector of a gray image that corresponds with an original colorful image, a gradient image and an adaptive gain function; transforming the original color image from an RGB spaceto an HIS space through nonlinear transformation; performing enhancement processing on a brightness vector in the HIS space image by means of a contrast limited adaptive histogram equalization algorithm; transforming the enhanced HIS space image to the RGB space; performing generalized bounded multiplication operation based on an adaptive gain function on an R component, a G component and a B component in the enhanced RGB image, thereby acquiring the enhanced image based on the gradient information of the original image; performing image displaying after enhancement; and performing quantitative evaluation on the enhanced image. The underwater target detecting image enhancement method can sufficiently use the texture of the original image for realizing image enhancement processing, therebyimproving visual quality of the processed image and obtaining abundant gradient information.

The invention relates to the field of machine vision and image processing, in particular to a method for attaching a mobile phone cover lens to a liquid crystal display .The method comprises the steps of detecting and positioning the angular point of the cover lens through Canny edge detection and Hough transformation, detecting and positioning the angular point of the liquid crystal display through wavelet algorithm, Sobel edge detection and Hough transformation, calculating the actual three-dimensional coordinates of the cover lens and the liquid crystal display through coordinate system transformation based on the stereoscopic parallax principle, and then adjusting the cover lens and the liquid crystal display according to the three-dimensional coordinate difference value of the two to coincide in the X-Y plane, so that automatic attaching of the cover lens and the liquid crystal display is achieved under visual guidance .By the adoption of the method, automatic attaching of the cover lens and the liquid crystal display is achieved, attaching speed is high, efficiency and precision are high, processes are simple, no personnel limitation is set, manpower resource is saved a lot, cost is low, and adaptability is high.

The invention discloses a cable apparent image defect detection device, which comprises a zynq processing platform and an off-chip DDR3 memory, wherein the zynq processing platform comprises an ARM part, an FPGA hardware processing part and a bus interconnecting part. The cable apparent image defect detection device is used for detecting wire apparent defects, adopts a high-speed camera for imaging cable apparent, is connected to a system through a camera link interface, and finally sends defected images after marking to an HDMI display for displaying results through six image processing steps including median filtering, sobel edge detection, threshold value segmentation, morphological filtering, boundary tracing and target recognition. The cable apparent image defect detection device implements parts of image processing algorithms including median filtering, sobel edge detection, threshold value segmentation and morphological filtering on hardware, and substantially improves the image processing speed.

The invention discloses a method for extracting foreground images. The method comprises the steps of classification of the field depth of images and segmentation of color information. According to the step of the classification of the field depth of the images, field depth areas of the foreground images, field depth areas of background images and a judgment area where the field depth is not detected are found out by adopting a Kinect device. The method for segmenting the color information includes the first step of converting the images from RGB color space to L*a*b color space, the second step of performing Sobel edge detection on the images, the third step of performing watershed segmentation on edge gradient values obtained from the second step, the fourth step of allowing the L*a*b color information to correspond to the blocks after the watershed segmentation and performing the computation of average values and the computation of the standard deviation, the fifth step of comparing the characteristic values of foreground image areas and the characteristic values of the judgment area, adding the similar areas into the foreground image areas and adding the dissimilar areas into the background images. According to the method for extracting the foreground images, the Kinect classification area is supplemented and corrected by combining the color characteristics of the images, the precision rate of the Kinect device for extracting the foreground images is improved, meanwhile, the operation time for extracting is shortened, and computational efficiency is enhanced.

The present invention discloses a matching method for a vehicle track shielded by an overpass in an aerial video. The matching method comprises the following steps: (1) detecting the border of the overpass, confirming the area of the overpass and confirming the edge of the overpass by utilizing Sobel edge detection combined Hough Transfom; (2) extracting histogram feature in gradient orientation of a tracked vehicle as the template feature of the vehicle; (3) estimating the time interval from the moment that the vehicle comes into the overpass to the moment that the vehicle leaves the overpass according to the driving speed of the tracked vehicle; (4) detecting the corresponding tracked vehicle at the other side of the overpass. In the matching method, the leaving area of the vehicle is cleared through the detection for the area of the overpass, so as to shrink the detection range of the vehicle leaving the overpass and shorten the detection time; meanwhile, the obtained overpass width data can be used for estimating the time that the vehicle leaving the overpass.

The invention relates to an efficient medical image segmentation method based on a game framework, and belongs to the field of computer vision. The method adopts a game framework model, the Sobel edge detection algorithm and the Otsu threshold adaptive algorithm for image operation, and includes sample set establishment, image pre-operation, candidate point set selection and detailed segmentation. By means of the method, duplicate-removal calculation of the candidate point marking process on a physical model and optimized selection of the candidate point set can be performed according to a medical image sample set, and the payoff matrix storage space is saved in the segmentation process based on the game framework; according to experiments, segmentation time is shortened, memory space requirements of the algorithms are lowered, and segmentation accuracy is improved.

An edge-oriented interpolation method is used for enlarging a source image. The region in which each destination pixel in an enlarged image is located is identified as a smooth region, a text region or an edge region, and Sobel edge detection for two adjacent pixels around the destination pixel is performed to obtain two Sobel gradients. When the destination pixel is located in a smooth region, cubic convolution interpolation is performed for the destination pixel with a first slope. When the destination pixel is located in a text region, cubic convolution interpolation is performed for the destination pixel with a second slope. When the destination pixel is located in an edge region, Sobel-based interpolation for the destination pixel is performed with the two Sobel gradients.

The invention discloses an airfield runway foreign matter and crack recognizing system and a detecting and recognizing method thereof. The system comprises a camera, an FPGA processor and a DSP which are sequentially connected. The FPGA processor and the DSP are connected with a display module. The method includes the steps that S1, the camera collects a runway color image of an airfield runway pavement; S2, the FPGA processor reads the runway color image and converts the runway color image into a runway gray image for Sobel edge detection; S3, after edge detection, one path is stored in an SDRAM, and Hough conversion is conducted on the other path for removing graticule; S4, morphological filtering is conducted; S5, whether foreign matter and cracks exist in the runway image or not is detected, and if yes, a runway edge detection image is sent to the DSP to be processed; S6, the edge detection image is filled to form characteristics; S7, the characteristics are extracted and foreign matter or cracks are recognized.

The invention discloses a convolutional neural network method and support vector machine method-based image identification method for an imaging department. The method comprises the following steps ofcollecting an original medical image as a sample, and performing grayscale processing on the original medical image by using a weighted grayscale algorithm to obtain a grayscale image; processing thegrayscale image by using histogram equalization to obtain an equalized grayscale image, performing edge detection on the equalized grayscale image by using an improved Isotropic Sobel edge detectionoperator to obtain an edge image, and performing binarization on the edge image by using an adaptive threshold algorithm to obtain a binarized medical image; processing the binarized medical image byusing morphological arithmetic operation to obtain a medical candidate region image, forming training data by the candidate region image, initializing a convolutional neural network, inputting the training data to the convolutional neural network, and extracting feature data; and transmitting the feature data extracted by training a convolutional neural network model to a support vector machine for performing training, and inputting extracted test feature data to a medical image identification training model for performing judgment, thereby finally obtaining an accurate medical image identification result.

The invention discloses a method and a device for recognizing a video format, equipment and a storage medium. The method comprises: acquiring a target image in a video to be played; sequentially usingat least one image recognition algorithm to process the target image according to a preset execution sequence to obtain a processing result; determining a format of the video to be played according to the processing result, wherein the image recognition algorithm comprises a Sobel edge detection algorithm, a histogram statistical algorithm, an image gray processing algorithm, an image Gaussian filtering algorithm, an image contour extraction algorithm, a graph Hu invariant moment comparison algorithm and a multi-region central moment algorithm. The technical solution provided by the embodiments of the invention realizes the technical effects of automatically recognizing all the 3D video formats, using at least one image recognition method to recognize the image in the process of recognition, improving the accuracy of the determination result of the video format, and saving labor cost.

The invention discloses a method for measuring thickness of a cable protecting bush accurately, and solves the defects that the accuracy is low, the error is large, and the efficiency is low by using a traditional geometrical measurement. The method includes the steps that firstly, preprocessing is conducted on a collected slice image of the cable protecting bush, edge pixel coordinates are obtained by using an Sobel edge detection method and a contour tracing technology, detection and distinguishing are conducted on angular points based on gradient correlation matrixes, as for the angle points, sub pixel position is conducted by using a support line fitting method based on angle point neighborhood supporting points, as for non-angle points, accurate positioning of the non-angle points is completed by using an edge re-positioning algorithm based on ZOM sub pixels. By means of the method, the measuring accuracy of the cable protecting bush is further improved, and the method is easy to carry out.

The invention relates to a self-adaptation threshold value video streaming multi-texture-direction error concealment method which is characterized in that based on an improved self-adaptation threshold value Sobel edge detection algorithm, interpolation directions of pixel points in error macro blocks are determined in a self-adaptation mode by detecting possible edge directions of adjacent blocks of the error macro blocks in cooperation with a boundary pixel difference value cost function, then the direction interpolation is carried out on the pixel points in the error macro blocks to recover the error macro blocks, edge information can be accurately extracted, lost macro blocks can be recovered in a self-adaptation mode, and the smoothness of recovered video images can be guaranteed. Compared with an H.264 standard error concealment algorithm, the self-adaptation threshold value video streaming multi-texture-direction error concealment method has the advantages that both subjective vision judgment and objective numerical calculation of the video image concealment effect are improved to a certain degree.

The invention discloses a town information extraction method based on a Google Earth remote sensing image. The method is characterized by, to begin with, performing pretreatment, such as median filtering and the like, on the remote sensing image; then, generating an edge image of the image based on a sobel edge detection operator; next, constructing a Gauss spatial domain weighting edge density generation operator and generating an edge density image; carrying out OSTU optimal threshold segmentation on the edge density image; and at last, carrying out aftertreatment on the segmentation image, and finally obtaining town information of the remote sensing image. The extraction method in the invention is good in effect, fast in speed, and high in accuracy, and is suitable for information extraction of towns, especially towns at the connection places of cities and suburbs.

The invention discloses a commodity falling detection method of an automatic vending machine based on double cameras. The method comprises detection steps that (1) when a facility detects that a usercompletes payment, a holder rises to a layer where a commodity is positioned and is stopped, and the cameras at the two sides of the holder are started for snapshot; (2) a local processing module of the facility is started to process each snapshot image cutting the snapshot image according to a pre-determined monochrome chequer position of a side plate of the holder, removing irrelevant areas, graying the cut image, carrying out edge detection on the grayed image via a sobel edge detection operator to generate an edge image, and storing the edge image as a background image; (3) after that thebackground is shot, a commodity channel conveyor belt is started, and the cameras at the two sides of the holder are started to shoot the state of the holder; and (4) the conveyor belt is stopped andthe cameras are switched off. The method can be used to detect small, flat and transparent commodities, and influence of the ambient light on commodity detection can be avoided effectively.

The invention relates to a face detection method and system. The method comprises the following steps: acquiring a to-be-detected original image; carrying out skin color detection and foreground detection on an original image to extract a suspected face area in the original image, and drawing a target box for the suspected face area; defining the size of a target frame, and screening the target frame by adopting a Sobel edge detection algorithm to filter out non-target frames in the target frame; sequentially performing R-Net operation and O-Net operation are on the filtered target frame to obtain a target frame including a real face region and five feature points in each of the target frames for indicating a face contour.According to the technical scheme, a traditional MTCNN network is improved, a skin color detection method and a foreground detection method are combined, a suspected face area is extracted, the operand is greatly reduced under the condition that the frame scale of anoriginal image is not reduced, the detection precision and the operation time are balanced, the cost of face detection equipment is reduced, and higher universality is provided.

The invention discloses a real-time image segmentation processing system and a high-speed intelligent unified bus interface method based on an Institute of Electrical and Electronic Engineers (IEEE) 1394 interface, which are used for solving the technical problem that the speed of the real-time segmentation processing which is carried out on camera image data streams of an IEEE 1394 camera by using an existing image processing system is low. The technical scheme is as follows: an IEEE 1394 controller is designed to analyze the IEEE 1394 bus protocol of the high-speed camera; effective image data are received in accordance with image frame synchronizing signals; a parallel image processing hardware structure is built in an field programmable gate array (FPGA); Sobel edge detection, threshold segmentation and morphology corrosion processing are adopted to realize the real-time segmentation processing on the high-speed image data streams; the processed result images are forwarded at a high speed by using a high-speed receiver SerDes through fiber channels; a multilevel hardware pipeline design is used to quicken to process system images; and a clock control module is used for realizing the switch between an IEEE 1394 bus synchronous clock and a high-speed intelligent bus synchronous clock, thus realizing high-speed and reliable transmission for image data of the two types of buses.

The invention provides a residual filtering method based on a pixel domain JND model. The method comprises the following steps: firstly, performing modelling in a pixel domain to obtain a JND threshold of each pixel, then, mining the correlation between the JND threshold of each pixel and the JND thresholds of neighboring pixels in a conversion unit, performing block classification on the conversion unit by using a Sobel edge detection operator to obtain a weight factor of the conversion unit, and finally, filtering the residual in combination with the spatial correlation of the JND thresholds of the pixels and the weight factor of the conversion unit. The proposed method model can be embedded in an HEVC video coding framework. The experimental results show that in the case that the subjective perceived quality of human eye does not decrease significantly, more actual distortion can be tolerated, and a certain code rate can be reduced.