414results about How to "Shorten recognition time" patented technology

The invention discloses a user online authentication method and system based on living body detection and face recognition. The method includes the user online registration step and the user online authentication step. The user online authentication step comprises the living body detection step, the image processing step, the feature value extraction step, the face comparison step and the result processing step. In the living body detection step, whether an authenticated user is a living body or not is determined and a face picture is acquired. In the image processing step, the collected face picture is processed. In the feature value extraction step, face part features of the processed face picture are extracted. In the face comparison step, extracted feature data of the collected face image are compared with corresponding face data in a user face feature value database, a threshold value is set, and when similarity exceeds the threshold value, the acquired result through matching is output. The user online authentication method and system can avoid authentication cheating through videos including faces, safety of the system is improved, recognition time can be shortened, and recognition accuracy is improved.

The invention relates to a mobile phone casing defect detecting method based on depth learning. The method comprises the steps that (1) the image of a mobile phone casing to be detected is acquired and pre-processed; and (2) the pre-processed image is input into a pre-trained defect detection model for defect detecting to acquire the position of a defect on the mobile phone casing, and the confidence of the position as the defect is provided. The defect detection model is a depth network based on depth learning, and comprises a feature extraction network and a classifier and regression device network, wherein the feature extraction network and the classifier and regression device network are in successive cascade. The feature extraction network carries out feature extraction on the pre-processed image to acquire a feature image. The classifier and regression device network classifies and regresses the feature image to acquire the defect position and the confidence of the mobile phone casing. Compared with the prior art, the method provided by the invention has the advantages of high detection precision and accurate and reliable detection result.

The invention provides a face attribute recognition method based on the multi-task deep learning and relates to the face attribute recognition technique in the field of computer vision. The method comprises the steps of preparing an image data set; subjecting each image in the image data set to face detection one by one; detecting face key points in all detected faces; aligning each face with a standard face image according to the face alignment method based on detected face key points to form a face image training set; calculating an average face image in the training set; constructing a multi-task deep convolutional neural network, and training network parameters after subtracting the average face image from each face image in the face image training set so as to obtain a convolutional neural network model; detecting faces and face key points in a to-be-recognized test image, and aligning each face in the above image with the standard face image based on the face key points; placing the standard face image into the constructed convolutional neural network model after subtracting the average face image from the standard face image and conducting the feedforward arithmetic operation so as to obtain a result.

The invention discloses a three-dimensional convolutional neural network (3D CNN) based video classifying method and belongs to the technical field of video processing. According to the method, a video is sampled at equal intervals to obtain a plurality of video segments, a video database is amplified, three-dimensional video segments are directly input into a 3D CNN, and time domain and space domain characteristics of the video are extracted, so that the limitation of a conventional video classifying method in manually selecting video characteristics and video modeling modes is improved. A parallel distributed 3D CNN multi-classification model lowers the complexity in learning the 3D CNN and enables a classification system to realize distributed parallel computation more conveniently. Relatively high identification rate can be achieved with only fewer video segments based on a 3D CNN multi-classification system, and videos not belonging to any type can be classified into new type, so that the classification error of the new type is avoided.

The invention relates to a dispatching management system and a method for event emergency response of urban bus passenger transport. Methods of combining actual measurement data, event prediction model and algorithm as well as event real-time dispatching model and algorithm are comprehensively applied; geographical information technology, locating technology and modern communication technology are scientifically integrated; bus emergency dispatching technology developed through bus network intellectualization and multi-way coordinative dispatching technology of an urban bus passenger transport system are utilized; and an emergency coordinative dispatching model and the method are established by taking the networked bus passenger transport vehicle dispatching and operation dispatching management as a core and based on the key theories and methods for multi-way and multi-line intelligent coordinative dispatching of a urban bus and passenger transport system. The invention provides excellent theory, methods and technique supports for the real-time dispatching of urban bus events, effectively reduces the event recognition time, generates a quick real-time dispatching strategy, effectively eliminates the influences generated by the events, and resumes the driving plan so as to effectively improve the running efficiency and safety of buses.

The invention discloses a live face identification method. The method comprises steps that video frame images shot by a camera are acquired; whether the video frame images are live face images can be identified through a trained convolutional neural network. The invention further discloses a live face identification system, the system comprise a camera module, an acquisition module and a live face detection module, wherein the acquisition module is connected with the camera module and the live face detection module, the acquisition module is used for sampling videos shot by the camera module to acquire the video frame images, and the live face detection module is used for detecting and identifying whether the video frame images are live face images through the trained convolutional neural network. The method and the system are advantaged in that brains of human beings are simulated through employing a depth learning neural network, whether the video frame images are live face images is identified through the trained convolutional neural network, whether the images are live faces or pictures can be identified through intuition of a computer to a great degree but not a designed algorithm, so algorithmic defects can be avoided, and identification accuracy is substantially improved.

The invention belongs to the computer vision and intelligent identification technology field and especially relates to a cow face identification method based on a convolutional neural network and a classifier model. A last hidden layer of the convolutional neural network is a fully-connected layer containing 32, 64, 128, 256 or 512 nerve cells and is used for extracting a characteristic. And then, the classifier model is used to complete identification of a cow individual. When there is a newly-added cow, image data of the cow only needs to be collected, the data is input into a convolutional neural network model, and a characteristic is extracted and is added to an original classification model so that identification can be performed; and the convolutional neural network model does not need to be retrained. In the invention, the convolutional neural network model of a 64-dimension characteristic extraction layer is selected, sparsity is used to express a classification model, training data 24000 pictures and test data 6000 pictures of 30 cows, which are randomly selected, are tested, and a result shows that identification time is shortened through using the method; average time consuming for identifying each cow is shortened to 0.00022s; and an identification rate reaches more than 99%.

A cityscape image semantic segmentation method based on multi-feature fusion and Boosting decision forest includes the following steps of carrying out super-pixel segmentation on images, carrying out multi-feature extraction, carrying out feature fusion and carrying out training learning and classification recognition. The method effectively integrates 2D features and 3D features and remarkably improves recognition rates of targets. Compared with the prior art, segmentation results are consistent, connectivity is good, edge positioning is accurate, a Boosting decision forest classification mechanism is introduced, and stability of target classification is guaranteed.

Disclosed is a multi-camera system target matching method based on a deep-convolution neural network. The multi-camera system target matching method based on the deep-convolution neural network comprises initializing multiple convolution kernels on the basis of a local protective projection method, performing downsampling on images through a maximum value pooling method, and through layer-by-layer feature transformation, extracting histogram features higher in robustness and representativeness; performing classification and identification through a multi-category support vector machine (SVM) classifier; when a target enters one camera field of view from another camera field of view, performing feature extraction on the target and marking a corresponding target tag. The multi-camera system target matching method based on the deep-convolution neural network achieves accurate identification of the target in a multi-camera cooperative monitoring area and can be used for target handoff, tracking and the like.

The invention discloses a multi-task deep convolutional neural network-based vehicle color identification system. The system comprises a high-definition camera mounted above lanes on a road, a trafficcloud server and a vehicle color visual detection subsystem; the vehicle color visual detection subsystem comprises a vehicle locating detection module, a license plate locating detection module, a license plate background color identification module, a color difference calculation module, a vehicle color correction module and a vehicle color identification module; the vehicle locating detectionmodule, the license plate locating detection module and the vehicle color identification module share a same Faster R-CNN deep convolutional neural network; by adopting the deep convolutional neural network, vehicles on the road are quickly segmented; license plates on the road are quickly segmented by further adopting the deep convolutional neural network through using vehicle images; and space position information of the vehicles and the license plates in a road image is given. The multi-task deep convolutional neural network-based vehicle color identification system provided by the invention is relatively high in detection precision and relatively high in robustness.

The invention provides a touch control equipment control method and device and touch control equipment. The touch control equipment control method includes the steps of detecting at least two pressure values of touch operations on a touch screen and controlling the touch control equipment to carry out a preset pressure triggering event when the at least two pressure values comprise at least two pressure values in different pressure ranges. According to the technical scheme, the mis-operation problem can be effectively solved, a user can conveniently operate the touch control equipment and user experience is effectively improved.

The invention provides a self-adaption endpoint detection method using a short-time time-frequency value and relates to a voice detection technology in a speaker recognition system. The self-adaption endpoint detection method comprises the following steps: after inputting a voice signal, analyzing a voice file and extracting a sampling value; pre-processing an obtained voice sampling sequence; dividing a pre-processed signal into frames with fixed lengths to form a frame sequence; aiming at data of each frame, extracting three voice signal characteristic parameters of relative values of short-time energy, short-time information entropy and a short-time range; calculating the short-time time-frequency value of each frame of the signal according to the three voice signal characteristic parameters to form a short-time time-frequency value sequence; analyzing a short-time time-frequency value sequence from the first frame of the signals, and finding a starting point and a finishing point of voices and outputting a voice endpoint detection result. The starting point and the finishing point of the voices can be accurately detected under complicated background noises; the recognition accuracy of the system is improved, the recognition time is shortened and the performance of the speaker recognition system under a complicated environment is improved.

The invention discloses a method for preventing collisions among a plurality of tags based on grouping dynamic frame and binary tree search in an RFID system. The method comprises a tag quantity estimation stage and a tag identification stage, wherein the tag quantity estimation stage is to finish the estimation of the quantity of unidentified tags; the tag identification stage is to select the optimal grouping number and the optimal frame size of each group according to the estimated value of the quantity of the unidentified tags, distribute the tags into a plurality of groups of frame periods for identifications in turn, and identify the tags in each group of time slots where collisions happen through the binary tree search so as to identify all the tags. The method combines the advantages of an ALOHA algorithm and a binary tree algorithm so that the quantities of collision time slots at the early stage of the identification and idle time slots at the late stage of the identification are greatly reduced, and the method has the advantages of simple structure, fast identifying speed and low tag power consumption, thus the method is quite suitable to be applied in the RFID system.

The invention discloses a SAR target recognition method based on a sparse least squares support vector machine, which belongs to the technical field of image processing and mainly solves the problem that the existing method need a long time for SAR target recognition. The realization process comprises the following steps of: firstly respectively implementing feature extraction to the selected target images with known classification information and images to be recognized to obtain training samples and test samples; and then applying iterative training to the training samples by using the combination of incremental learning method and reversal learning method to select a sparse support vector set and obtain a Lagrange multiplier and deflection corresponding to the support vectors in the set; and finally using a classification decision function to recognize the test samples according to the obtained support vector set, the Lagrange multiplier and deflection corresponding to the support vectors. The invention has the advantage of shortening recognition time under the condition of equivalent recognition precision and can be used for detection and recognition of SAR target.

The invention discloses a certificate picture identification method and belongs to the technical field of picture processing. The identification method comprises the following steps: marking original pictures of different types of certificates and at least one type of reshot pictures of the original pictures, and then, carrying out training through a convolutional neural network to obtain a classification model; and inputting a certificate picture to be identified to the classification model to obtain an identification result. The invention also discloses a certificate picture identification system. According to the certificate picture identification method, training is carried out through the convolutional neural network to obtain the corresponding classification model, and then, the certificate picture to be identified is input to the classification model to obtain the identification result to judge whether the certificate picture to be identified is an original picture or a reshot picture; and the method not longer needs manual analysis of the characteristics of the pictures, thereby reducing certificate picture identification time and enhancing identification effect.

The invention discloses a method for identifying an image with a watermark and an identifying system. The method includes the steps of selecting all candidate watermark areas of an image to be identified, identifying watermarks on each candidate watermark area through a deep convolution neural network classifier to determine whether the image to be identified is an image with a watermark, and realizing identification of images with watermarks. Lots of image training data can be conveniently and rapidly obtained, and a deep convolution neural network classifier is established through a convolution neural network algorithm via the image training data to solve the problem of insufficient training data in the prior art. The deep convolution neural network classifier established in the invention effectively simulate a visual processing system of human eyes, can identify local fine watermark texture, and effectively solves the problem of small area, light color and high transparent level of watermark in an image with a watermark. A process for identifying areas without watermarks can be reduced, the identification time is shortened, and identification efficiency is improved.

The invention relates to the field of image processing and image identification, and particularly relates to a human motion local characteristic representation method and an application in behavior identification thereof. The human motion local characteristic representation method comprises the following steps that firstly kinetic energy of human bone joints, the position coordinates of the human bone joints, the direction change vectors of the human bone joints and posture potential energy of the human bone joints are extracted from the perspective of energy according to human behavior biological and kinematic characteristics so that a local characteristic combination matrix is constructed; and then dimension reduction is performed on the local characteristic matrix by utilizing K-means clustering and bag of word (BOW) characteristics are extracted, characteristic vectors used for behavior identification are formed through combination of human joint angle characteristics, and the characteristic vectors are applied to the field of human behavior identification so that the method has great effect through experimental verification.

The invention provides an SVM (support vector machine) classifier training sample acquiring method, a training method and a training system. The SVM classifier training sample acquiring method includes calculating to acquire distance of each sample of an SVM classifier; according to the distance of each sample, clustering the samples for the first time to acquire at least one first category and the samples contained by each first category; clustering the samples for the second time to acquire at least one second category and the samples contained by each second category, wherein a second distance threshold value is larger than a first distance threshold value; dividing the samples in the second categories containing one sample only into isolated samples; selecting one sample in each first category as a representative sample, and setting the training samples of the SVM classifier according to the representative samples and the isolated samples. By the SVM classifier training sample acquiring method, the training method and the training system, number of the samples can be reduced effectively, complexity of sample space of the classifier is lowered, and classifier training is enabled to be simpler and more effectively.

The invention discloses a method for identifying a triangular star map, and belongs to the field of navigation, guidance and control of spacecraft. The method comprises the following steps of: 1, establishing a navigation database, namely constructing characteristic triangles, calculating a plane unit normal vector, solving an optimal projection principal axis and projection point values, and normalizing magnitude; 2, forming an observation triangle; and 3, searching a characteristic triangle which is most matched with the observation triangle from the navigation database, sequentially retrieving a projection point value block index stable, a characteristic triangle table and a magnitude normalizing table in a hierarchical retrieval mode, and finishing the matching of the observation triangle. The method for identifying the triangular star map solves the problems of low noise robustness, more redundant matching and low identification rate of the conventional method for identifying the triangular star map, obviously improves the search speed of the navigation database, reduces redundant matching, improves the identification rate, and has higher noise robustness.

The invention discloses a distributed tracking and monitoring method and system. The monitoring method includes the step of central control, the step of selecting a specific video camera and the step of monitoring through the specific video camera, wherein in the step of central control, a central service station sends a control instruction to a video camera group and receives feedback information of the video camera group; in the step of selecting the specific video camera, the central service station selects at least one video camera from the video camera group to serve as the specific video camera according to the feedback information and performs linkage monitoring; in the step of monitoring through the specific video camera, the specific video camera takes the initiative to recognize, track, report and monitor a target to complete continuous tracking and monitoring of the monitored target according to the control instruction of the central service station. To overcome the defects in the prior art, the distributed tracking and monitoring method and system can solve the problems that in an ultra-high definition image, a mobile target is large in recognition and calculation amount and difficult to track, the monitoring efficiency can be greatly improved particularly for the situation that the mobile target is tracked all over the city, and the distributed tracking and monitoring method and system has important significance for fast tracking and monitoring the target in real time.

The embodiment of the invention provides an image recognition method and a device. The method comprises the following steps: obtaining an image to be recognized, dividing the image to be recognized into a plurality of sub-images according to a preset size, and obtaining a first gray feature vector corresponding to each sub-image; acquiring a second gray feature vector corresponding to the matchingtemplate, calculating similarity between the sub-image and the matching template according to the first gray feature vector and the second gray feature vector; the sub-images whose similarity is greater than the preset threshold are selected as the target sub-images, and the target sub-images are recognized by the classification model according to the target sub-images. The apparatus is configured to perform the method. An embodiment of the present invention calculates a similarity between a sub-image and a matching template according to a first gray feature vector and a second gray feature vector, and the target sub-images whose similarity is greater than the preset threshold are selected as the target sub-images, and then the target sub-images are input into the classification model forrecognition, thereby reducing the recognition time and improving the recognition efficiency of the images to be recognized.

The invention relates to a hand-held laser code-jetting character recognizer under a complex background, which consists of to components including a hardware structure and a software structure, wherein the hardware structure comprises an image acquisition unit, an image processing unit, a data storage unit, a liquid crystal display unit, a GPRS communication unit, a key operation unit, a power supply module and the like; and the software structure comprises a system scheduling module, a hardware control module, a driver module, a hardware platform, a human-computer interface module, a database module, a recognition algorithm module and the like. By adopting an embedded type design, the hand-held laser code-jetting character recognizer integrates the image acquisition function, the digital image recognition function and the GPRS-based data transmission function, and can perform image acquisition and quick automatic recognition on an object on site and send the recognition result to an appointed sever through the GPRS communication unit. The recognizer is simple to operate and convenient to use, and has an extensive practical value and an application prospect in the technical field of pattern recognition detection devices.

The invention relates to a method and a system for identifying the posture of a body. The method comprises the following steps of: A, respectively acquiring motion parameters of a plurality of parts of the body and sensing parameters of hardware equipment; B, according to an angle difference value which is obtained by comparing the motion parameters of all parts of the body, acquiring the position change information of the body; C, according to the position change information of the body and the angle difference value of the sensing parameters of the hardware equipment, determining the motion information of the body relative to the hardware equipment; and D, identifying the motion information of the body relative to the hardware equipment, and outputting the motion information which serves as monitoring information. In the method for identifying the posture of the body provided by the invention, a sensor device is used for identifying the posture of the body, and simultaneously optical equipment can be externally connected for auxiliarily monitoring the position of the body relative to the hardware equipment. By adoption of the method and the system for identifying the posture of the body provided by the invention, the speed for identifying the posture of the body is improved, different parts of the body can be identified at the same time, and certain flexibility is realized.

The invention discloses a method and a system for identifying dynamic objects. The method includes distributing n video files to k distributed type deployment computing nodes so as to subject the n video files to object identification by the k computing nodes, receiving the n video files subjected to the object identification from the k computing nodes, and processing results of the n video files subjected to the object identification, wherein the k and the n are natural numbers, and the k is not larger than the n. By the method and the system for identifying dynamic objects, one one hand, for massive video data to be analyzed, enough computing nodes for processing can be deployed, and processing capacity is much higher than that of a single server, and on the other hand, the distributed type computing nodes can quickly identify the dynamic objects in the video files owning to a parallel processing manner, identifying time is greatly shortened, and effective utilization rate of video resources is increased.

The invention relates to an electric product which comprises an identification device capable of obtaining information related to food to be added in a storage chamber, a displaying portion for displaying information related to storing food in the storage chamber, a storage portion for storing information related to storing food in the storage chamber and a control portion for controlling the displaying portion. If the identification device obtains information related to food to be added in a storage chamber, the control portion stores the related information to the storage portion and displays the related information by the displaying portion.

The invention provides a dimensionality-reducing HOG-MBLBP fusion feature based traffic sign recognition method based on PCA.. The method of the invention comprises the following steps of: training aclassifier model by using a training sample; constructing a training sample database; extracting a training sample image of the determined training set for graying, extracting HOG features and MBLBP features; serially connecting two eigenvectors of HOG and MBLBP to obtain a HOG-MBLBP fusion eigenvector; the obtained fusion eigenvector being dimensionally reduced by using a PCA algorithm; a linearsupport vector machine (SVM) algorithm being used to train the obtained dimension-reduced fusion eigenvector to obtain an SVM traffic sign classifier; obtaining a traffic sign image; traffic sign images being recognized by the classifier model. The technical proposal of the invention solves the problems that the traffic sign recognition method in the prior art has low recognition accuracy rate andlong operation time, and is difficult to meet the requirement of vehicle-mounted real-time performance.

The invention relates to a three-dimensional image photographing device, a three-dimensional image photographing method and a face recognition method. The three-dimensional image photographing devicecomprises a laser source, an image sensor and a data processing module. The laser source is used for transmitting laser. The image sensor is used for sensing visible light for generating a two-dimensional image signal of an objective scene and sensing the laser which is reflected from the objective scene for generating a depth image signal. The data processing module is used for acquiring the two-dimensional coordinate of the objective scene according to the two-dimensional image signal, and acquiring the depth coordinate of the objective scene according to the depth image signal, thereby obtaining the three-dimensional coordinate information of the objective scene.

The invention discloses a pirate application identification method and device, and belongs to the technical field of information processing. The method comprises the following steps: obtaining a first name index of a first application, obtaining a second application matched with the first name index in a prestored application set, wherein the application set is accumulated by legal copy applications or pirate applications identified when pirate application identification is carried out before; if at least one second application is obtained, matching the first application with each second application one by one; and according to a matching result, carrying out the pirate application identification of the first application. The second application matched with the name index of the first application in the prestored application set is obtained, the first application and each second application are matched one by one, the pirate application identification is carried out according to the matching result, so that the amount of the second applications matched with the first application one by one is reduced, identification time is shortened, and identification efficiency is improved.

The invention relates to an intelligent mobile terminal scene character processing method. The method comprises the following steps: step 1, performing text coarse detection based on edges; step 2, obtaining a stroke width graph T of an input scene image I, analyzing the stroke widths and geometric characteristics of each candidate text area in a candidate text area set S, rejecting a non-text area not according with requirements, and finally outputting a positioning result graph L1; step 3, performing identification pretreatment; step 4, performing extraction operation of standard and directional element characteristics on segmented single characters; and step 5, performing fine classification based on Gabor characteristics. Compared to the prior art, the method provided by the invention has the following advantages: the accuracy is greatly improved, the recall rate is quite high, the time performance is substantially improved, and the accuracy of character identification is greatly enhanced.

The invention discloses an express sheet recognition method, an express sheet recognition system, terminal equipment and a storage medium. The method comprises: obtaining an express sheet picture to be recognized by performing optical character recognition on the express sheet to be recognized; extracting features of the express sheet picture to be recognized by a full convolution network to obtain a target feature map; selecting a target positioning window from the preset candidate positioning window set by using a non-maximum value suppression algorithm; acquiring a current character image from the target feature map according to the target positioning window; substituting the current character image into the preset feature extraction model to obtain a character feature sequence, and inputting the character feature sequence to the cyclic neural network to obtain a character recognition result. The invention may accurately locate the text information in the express sheet to be recognized, and filter out the invalid information with low quality requirement for the express sheet picture as well as strong anti-interference ability, thereby reducing the workload of express sheet recognition, saving a large number of express sheet recognition time, improving the express sheet recognition speed, and improving the user experience.