2164 results about "Visual technology" patented technology

The present invention is a method and system for automatically analyzing a category in a plurality of the categories in a physical space based on the visual characterization, such as behavior analysis or segmentation, of the persons with regard to the category. The present invention captures a plurality of input images of the persons in the category by a plurality of means for capturing images. The present invention processes the plurality of input images in order to understand the shopping behavior of the persons with the sub-categories of the category and analyzes the level of engagement and decision process at the sub-category level. The processes are based on a novel usage of a plurality of computer vision technologies to analyze the visual characterization of the persons from the plurality of input images. The physical space may be a retail space, and the persons may be customers in the retail space.

The invention relates to a movement human abnormal behavior identification method based on template matching, which mainly comprises the steps of: video image acquisition and behavior characteristic extraction. The movement human abnormal behavior identification method is a mode identification technology based on statistical learning of samples. The movement of a human is analyzed and comprehended by using a computer vision technology, the behavior identification is directly carried out based on geometric calculation of a movement region and recording and alarming are carried out; the Gaussian filtering denoising and the neighborhood denoising are combined for realizing the denoising, thereby improving the independent analysis property and the intelligent monitoring capacity of an intelligent monitoring system, achieving higher identification accuracy for abnormal behaviors, effectively removing the complex background and the noise of a vision acquired image, and improving the efficiency and the robustness of the detection algorithm. The invention has simple modeling, simple algorithm and accurate detection, can be widely applied to occasions of banks, museums and the like, and is also helpful to improve the safety monitoring level of public occasions.

The present invention is a method and system for characterizing physical space based on automatic demographics measurement, using a plurality of means for capturing images and a plurality of computer vision technologies. The present invention is called demographic-based retail space characterization (DBR). Although the disclosed method is described in the context of retail space, the present invention can be applied to any physical space that has a restricted boundary. In the present invention, the physical space characterization can comprise various types of characterization depending on the objective of the physical space, and it is one of the objectives of the present invention to provide the automatic demographic composition measurement to facilitate the physical space characterization. The demographic classification and composition measurement of people in the physical space is performed automatically based on a novel usage of a plurality of means for capturing images and a plurality of computer vision technologies on the captured visual information of the people in the physical space. The plurality of computer vision technologies can comprise face detection, person tracking, body parts detection, and demographic classification of the people, on the captured visual information of the people in the physical space.

The invention is related to methods and apparatus that use a visual sensor and dead reckoning sensors to process Simultaneous Localization and Mapping (SLAM). These techniques can be used in robot navigation. Advantageously, such visual techniques can be used to autonomously generate and update a map. Unlike with laser rangefinders, the visual techniques are economically practical in a wide range of applications and can be used in relatively dynamic environments such as environments in which people move. One embodiment further advantageously uses multiple particles to maintain multiple hypotheses with respect to localization and mapping. Further advantageously, one embodiment maintains the particles in a relatively computationally-efficient manner, thereby permitting the SLAM processes to be performed in software using relatively inexpensive microprocessor-based computer systems.

The invention relates to a rapid target detection method based on a convolutional neural network, and relates to the computer vision technology. The rapid target detection method comprises the following steps: training convolutional neural network parameters by utilizing a training set; solving the problem of max-pooling losing feature by using an expander graph and generating a discriminative complete feature graph; regarding the full-connection weight of the convolutional neural network as a linear classifier, and estimating the generalization error of the linear classifier on the discriminative complete feature by using a probable approximately correct learning framework; estimating the required number of the linear classifiers according to the generalization error and the expected generalization error threshold value; and finally, completing the target detection on the discriminative complete feature graph by using the linear classifiers on the basis of a smooth window. The detection efficiency and the target detection precision are obviously improved.

The invention relates to a visual computing and prewarning method of deviation angle and distance of an automobile from a lane line. The image processing and computer vision technologies are utilized, and the deviation angle and distance of the automobile from the lane line are computed in real time according to the road surface image acquired by a vehicle-mounted camera, thereby estimating the line crossing time for safety prewarning. The method comprises the following steps: detecting the lane lines of the road surface image to obtain a linear equation of partial lane lines; establishing a three-dimensional coordinate system by using the camera as the initial point, and recording the mounting height and depression angle of the camera; calibrating the focal length according to the lane detection result under the condition of a given deflection angle; computing the deflection angle and vertical distance of the automobile relative to the lane line according to a pinhole camera model; and estimating the deviation time from the lane according to the instantaneous running speed of the automobile, thereby obtaining the safety prewarning or intelligent control information of the running automobile.

The invention belongs to the technical field of image processing and computer vision, and relates to a novel biomedical image automatic segmentation method based on a U-net network structure, including dividing a biomedical data set into a training set and a test set, and normalizing the test set and augmented test set; inputting the images of the training set into the improved U-net network model, and generating a classification probability map by output image passing through a softmax layer; calculating the error between classification probability diagram and gold standard by a centralized loss function, and obtaining the weight parameters of network model by a gradient backpropagation method; entering the images in the test set into the improved U-net network model, and outputting the image to generate a classification probability map through the softmax layer; according to the class probability in the classification probability graph, obtaining the segmentation result graph of theimage. The invention solves the problems that simple samples in the image segmentation process contribute too much to the loss function to learn difficult samples well.

A system and method for using computer image processing for selectable task-based digital video compression is described. The system is intended to reduce travel of experts and let these experts direct field agents in performing tasks remotely. The tasks to be performed in the remote field can be monitoring, manipulating, and navigating. A field agent performs the manipulation and navigation operations, this agent can be a human being or a robot. The task-based compression algorithms use computer vision techniques to extract the bare minimum amount of information from the remote field scenery to allow the task to be performed. High frame rate photo-realistic reconstruction of the remote scene is not generally necessary.

The present invention is a method and system for optimizing the observation and annotation of complex human behavior from video sources by automatically detecting predefined events based on the behavior of people in a first video stream from a first means for capturing images in a physical space, accessing a synchronized second video stream from a second means for capturing images that are positioned to observe the people more closely using the timestamps associated with the detected events from the first video stream, and enabling an annotator to annotate each of the events with more labels using a tool. The present invention captures a plurality of input images of the persons by a plurality of means for capturing images and processes the plurality of input images in order to detect the predefined events based on the behavior in an exemplary embodiment. The processes are based on a novel usage of a plurality of computer vision technologies to analyze the human behavior from the plurality of input images. The physical space may be a retail space, and the people may be customers in the retail space.

The present invention is a method and system for analyzing shopping behavior by associating RFID data, such as tracking data by the RFID tag identifications, with video-based behavior and segmentation data, such as behavior analysis and demographic composition analysis of the customers, utilizing a plurality of means for sensing and using RFID tags, a plurality of means for capturing images, and a plurality of computer vision technologies. In the present invention, the association can further comprise the association of the RFID with the transaction data or any time-based measurement in the retail space. The analyzed shopping behavior in the present invention helps people to better understand business elements in a retail space. It is one of the objectives of the present invention to provide an automatic video-based segmentation of customers in the association with the RFID based tracking of the customers, based on a novel usage of a plurality of means for capturing images and a plurality of computer vision technologies on the captured visual information of the people in the retail space. The plurality of computer vision technologies can comprise face detection, person tracking, body parts detection, and demographic classification of the people, on the captured visual information of the people in the retail space.

The invention discloses a wisdom skynet video behavior analyzing system which comprises a plurality of front end video collecting systems and a central monitoring system. The central monitoring system comprises a plurality of video behavior analyzing servers, a storage server, a monitoring server and a client side, wherein each video behavior analyzing server receives video information of one front end video collecting system respectively, automatically analyzes all kinds of exception behaviors by the adoption of an advanced image analyzing processing technology, a computer vision technology, a computer network technology and an artificial intelligent technology and generates alarming signals, the video behavior analyzing servers are respectively connected with the storage server, the monitoring sever and the client end through an interchanger, and the monitoring server is further connected with a third party alarm linkage server. The wisdom skynet video behavior analyzing system is provided with multiple detection modules, users can selectively install certain modules according to actual needs, installation is convenient, the operation is simple, and video behavior analysis can be carried out automatically. A distribution type structure is adopted by the wisdom skynet video behavior analyzing system, and data transmitting and processing speed can be improved.

The invention relates to the technical field of stereo vision, in particular to a stereo matching method. The method solves the problem that the accuracy of disparity correction optimization of the existing stereo matching method is insufficient. The stereo matching method based on disparity map pixel classification correction optimization comprises the following steps that (I) cost aggregation is conducted by taking a left view and a right view as references and based on a method combining a gray scale difference with a gradient, and a left disparity map and a right disparity map are obtained and subjected to left and right consistency detection to generate an initial reliable disparity map; (II) correlation credibility detection and weak texture area detection are conducted, and a pixel is classified into stable matching pixel points, unstable matching pixel points, occlusion area pixel points and weak texture area pixel points; (III) the unstable matching points are corrected by an adaptive weight algorithm based on improvement, and the occlusion area points and the weak texture area points are corrected by a mismatching pixel correction method; and (IV) the corrected disparity maps are optimized by an algorithm based on division, and dense disparity maps are obtained.

The invention relates to an unmanned aerial vehicle aerial photography sequence image-based slope three-dimension reconstruction method. The method includes the following steps that: feature region matching and feature point pair extraction are performed on un-calibrated unmanned aerial vehicle multi-view aerial photography sequence images through adopting a feature matching-based algorithm; the geometric structure of a slope and the motion parameters of a camera are calculated through adopting bundle adjustment structure from motion and based on disorder matching feature points, and therefore, a sparse slope three-dimensional point cloud model can be obtained; the sparse slope three-dimensional point cloud model is processed through adopting a patch-based multi-view stereo vision algorithm, so that the sparse slope three-dimensional point cloud model can be diffused to a dense slope three-dimensional point cloud model; and the surface mesh of the slope is reconstructed through adopting Poisson reconstruction algorithm, and the texture information of the surface of the slop is mapped onto a mesh model, and therefore, a vivid three-dimensional slope model with high resolution can be constructed. The unmanned aerial vehicle aerial photography sequence image-based slope three-dimension reconstruction method of the invention has the advantages of low cost, flexibility, portability, high imaging resolution, short operating period, suitability for survey of high-risk areas and the like. With the method adopted, the application of low-altitude photogrammetry and computer vision technology to the geological engineering disaster prevention and reduction field can be greatly prompted.

The invention discloses a target identification model training method, a target identification model training device, a target identification method, a target identification device and computing equipment, which belong to the technical field of computer vision. The target identification model training method comprises the steps of: inputting a plurality of local candidate regions selected for a training image into a target identification model, so as to obtain a preliminary result of classification of the plurality of local candidate regions output by the target identification model; performing local candidate region integration according to weak supervisory information and the preliminary result of classification of the plurality of local candidate regions; correcting parameters of the target identification model according to the preliminary result of classification of the plurality of local candidate regions and a result of local candidate region integration; and iteratively executing the training steps till a training result of the target identification model satisfies a predetermined convergence condition. The target identification model training method, the target identification model training device, the target identification method, the target identification device and the computing equipment can perform direct supervision of pixel level, can optimize a semantic segmentation model in an end-to-end manner, and can improve the result of target identification according to judgment of the local candidate regions.

The invention provides a simulation method and device of a digital twin system of an industrial robot. The simulation device comprises the industrial robot, a vision perception unit and a computer, wherein the vision perception unit is arranged at the tail end of the robot and is composed of a camera and a wire structure light emitter; an industrial robot simulation system is established, the robot is modeled, and three instructions are analyzed and run; a target three-dimensional measurement model of a robot visual system is acquired by using a wire structure optical geometric triangular method; a motion instruction is determined for the working task of the robot, virtual simulation is carried out through the computer, and reachable points and collision point detection are carried out; and the robot is actually used for identifying a target object through the visual perception unit, and the actual movement of the robot is driven. The simulation device is safe and reliable in virtual simulation, abnormal operation possibly in actual operation is avoided, the visual perception unit is used for identifying the target object through three-dimensional visual technology, the self-adaptive capacity of the robot to the field environment is improved, and flexibility and intelligence are enhanced.

The invention relates to an intelligent advertising system having functions of audience characteristic perception and counting and belongs to the technical field of computer vision. The system consists of a camera, computing equipment and an electronic advertisement screen provided with a memory, wherein the camera is responsible for acquiring an image and sending acquired image information to the computing equipment; and the computing equipment is a computer or an embedded microcomputer system which is responsible for performing processing and an operation on the image acquired from acquisition equipment and controlling the electronic advertisement screen according to an operation result so as to realize the memorizing, the displaying, the playing and the switching of an advertisement content. The system provided by the invention is simple in structure, low in cost and convenient in operation, so that the problem that the conventional billboard performs only one-way pushing without consideration of the information of current advertisement audiences is solved; the value of the billboard relative to other billboards can be obtained according to the function of counting, and extremely referable data is provided for an advertiser; and the speed and the correction rate of face detection and face tracking are improved by utilizing a background modelling technology.

The invention discloses a system and a method for measuring water level based on image vision, which is used for measuring large measuring range water level of places of reservoirs, hydropower station and the like. A CCD camera is arranged right above a floater device for acquiring videos; the videos are converted into digital image frames through a video A/D acquisition module; a floater image area in a dynamic image is influenced by distance changing between the floater and the CCD camera, wherein the distance changing between the floater and the CCD camera is caused by water level changing; the acquired floater image area in the videos is computed through an FPGA processing module, and a relative distance between the floater and the CCD camera is obtained for measuring the water level. In the invention, the water level can be measured indirectly through the floater area in the videos measured by the image vision technology without an expensive water level sensor, and the cost is low; by adopting a modularized design method, the invention has the advantages of simple installation, convenience for on-site measurement and system maintenance, and high adaptability.

The invention provides a face attribute recognition method of a deep neural network based on cascaded multi-task learning and relates to the computer vision technology. Firstly, a cascaded deep convolutional neural network is designed, then multi-task learning is used for each cascaded sub-network in the cascaded deep convolutional neural network, four tasks of face classification, border regression, face key point detection and face attribute analysis are learned simultaneously, then a dynamic loss weighting mechanism is used in the deep convolutional neural network based on the cascaded multi-task learning to calculate the loss weights of face attributes, finally a face attribute recognition result of a last cascaded sub-network is used as the final face attribute recognition result based on a trained network model. A cascading method is used to jointly train three different sub-networks, end-to-end training is achieved, the result of face attribute recognition is optimized, different from the use of fixed loss weights in a loss function, a difference between the face attributes present is considered in the invention.

The invention relates to the field of mechanical arms and computer vision technology. The binocular vision based object location grabbing method for a mechanical arm includes steps of 1, mounting a binocular vision camera and a mechanical arm; 2, correcting inner parameters of the binocular vision camera; 3, obtaining depth information of a target object; 4, establishing an image planar coordinatesystem and a binocular vision camera coordinate system; 5, calculating the spatial position of the target object in the mechanical arm coordinate system; 6, calculating the grabbing gesture of the mechanical arm; 7, sending control information. Compared with the prior art, the method has advantages that 1, time requirements for real time grabbing can be met since an object recognition method is simple and the calculation load is comparatively low; 2, a problem of operation failure, which occurs when fine difference exists between the gesture of the target object and an ideal gesture, of a mechanism system adopting a traditional demonstration system is avoided.

The invention belongs to the technical field of computer vision. A method for semantic segmentation of scene point cloud is provided, a framework of point cloud semantic segmentation model for large-scale dense scene based on depth learning technology is designed, For the input large-scale dense scene point cloud, it can transform the three-dimensional information of point cloud into two-dimensional information which can be processed directly by convolution without losing the information, and complete the task of point cloud semantic segmentation combined with image semantic segmentation technology. In this framework, we can effectively solve the semantic segmentation task of large-scale dense scene point cloud. The semantic segmentation result of the scene point cloud obtained by the method of the invention can be directly used in tasks such as robot navigation, automatic driving and the like. And this method is especially effective in non-synthetic natural scenes.

The present invention relates to a free multi-view real time three-dimension reconstructing system and a method thereof. The system consists of a CCD vidicon, a computer assembled with a multi-channel image collecting card and a plane demarcating template. Two concentric circles and eight tangential paths with prolonged lines crossing the circle center are drawn on the center of the plane demarcating template and are used for demarcating the vidicon parameter. In the technical proposal of the present invention, a reconstructing object is fixed at the center of the demarcating template; the template is held by hand to drive the reconstructing object to rotate freely within the range of the view and the depth of field of a vidicon at a fixed position; image collection and object reconstruction are processed synchronously, thereby being capable of finding concentrated missed view in time and displaying reconstruction result in real time. The three-dimension reconstructing method comprises the steps of image segmentation, full-automatic vidicon demarcation and the three-dimension reconstruction of object shape, etc. The three-dimension reconstructing system of the present invention is based on the computer vision technology, and the reconstructing method is based on the profile recovery shape and can realize free multi-view real time three-dimension reconstruction.

The invention relates to an unmanned aerial vehicle (UAV) small target detection method based on motion features and deep learning features, which belongs to the technical field of image processing and computer vision. The method includes the following steps: processing an input video data set through a video image stabilization algorithm to make compensation for the motion of a camera; analyzingdetected moving candidate target regions in images; dividing the video data set into two parts, and carrying out training by using a training data set to get an improved candidate region generation network model; generating a candidate target for the video images of a test set through a candidate region generation network based on depth features obtained from training; fusing the candidate targetregions; carrying out training by using the training data set to get a deep neural network model based on dual channels, and obtaining an identification result by using the model; and applying a target tracking method based on multilayer depth features to the identification result in the previous step to get the final position of a UAV. A UAV in a video image can be accurately detected, and thus,support can be provided for the subsequent research in fields related to UAV intelligent monitoring.

The invention discloses a natural interactive method based on three-dimensional gestures, the method utilizes a computer vision technology to obtain local features of a hand by foreground segmentation and fingertip detection, and the local features include fingertip position, palm contour, palm center position and the like. By adopting the stereoscopic vision technology, the hand features such as the fingertip position, the palm center position and the like are reconstructed in the three-dimensional space. The finger tip position, the palm center position and the like in the three-dimensional space are parameterized, and a three-dimensional interactive model based on points, lines and planes is defined, thus realizing various three-dimensional gestures in the three-dimensional space, such as fingertip clicking, fingertip squeezing, palm overturning, fingertip directing and the like. The method needs only two ordinary network cameras to meet the demands of real-time man-machine interaction.