821 results about "Frame difference" patented technology

A video quality evaluation method based on artificial neural network, which belongs to the field of computer digital video processing. The evaluation algorithm computes damage extent of an image by analyzing spatial characters (blur, entropy, blocking effect, frequency-domain energy analysis, saturation) and time behaviors (frame-to-frame differences). Evaluation effect of the algorithm can be effectively modified by using colour space saturation as one of the parameters for non-reference evaluation algorithm. The system is designed based on artificial neural network, therefore, training process and testing process of the network are included in realization of the algorithm. To the selected training sample (video image sequence), firstly extracts six parameters of the sample, then obtains expected output (subjective evaluation result) in training through subjective evaluation. Characteristic parameters of the training sample and the corresponding subjective evaluation result are inputted to the artificial neural network as training parameters. The experiment shows that evaluation result obtained by the objective evaluation system is highly-consistent with visual sense of human eyes.

The invention discloses a method and a system for authenticating a shielded face, wherein the method comprises the following steps: S1) collecting a face video image; S2) preprocessing the collected face video image; S3) performing detection calculation on the shielded face, evaluating a position of a face image by utilizing a three-frame difference method according to motion information of a video sequence, and further confirming the position of the face according to an Adaboost algorithm; and S4) performing authenticating calculation on the shielded face, dividing a face sample into a plurality of sub-blocks, performing shielding distinguishment on the sub-blocks of the face by adopting a SVM(Support Vector Machine) binary algorithm combined with a supervising 1-NN k-Nearest neighbor method, if the sub-blocks are shielded, directly abandoning the sub-blocks, and if the sub-blocks are not shielded, extracting a corresponding LBP (Length Between Perpendiculars) textural feature vector for performing weighting identification, and then using a classifier based on a rectangular projection method to reduce feature matching times. According to the method for authenticating the shielded face, the detection rate and the detection speed for the local shielded face are effectively increased.

A system and method for effectively encoding and decoding electronic information includes an encoding system with a tiling module that initially divides source image data into data tiles. A frame differencing module then outputs only altered data tiles to various processing modules that convert the altered data tiles into corresponding tile components. A quantizer performs a compression procedure upon the tile components to generate compressed data according to an adjustable quantization parameter. An adaptive entropy selector then selects one of a plurality of available entropy encoders to most effectively perform an entropy encoding procedure to thereby produce encoded data. The entropy encoder may also utilize a feedback loop to adjust the quantization parameter in light of current transmission bandwidth characteristics.

The present invention discloses a surveillance video pedestrian re-recognition method based on ImageNet retrieval. The pedestrian re-recognition problem is transformed into the retrieval problem of an moving target image database so as to utilize the powerful classification ability of an ImageNet hidden layer feature. The method comprises the steps: preprocessing a surveillance video and removing a large amount of irrelevant static background videos from the video; separating out a moving target from a dynamic video frame by adopting a motion compensation frame difference method and forming a pedestrian image database and an organization index table; carrying out alignment of the size and the brightness on an image in the pedestrian image database and a target pedestrian image; training hidden features of the target pedestrian image and the image in the image database by using an ImageNet deep learning network, and performing image retrieving based on cosine distance similarity; and in a time sequence, converging the relevant videos containing recognition results into a video clip reproducing the pedestrian activity trace. The method disclosed by the present invention can better adapt to changes in lighting, perspective, gesture and scale so as to effective improve accuracy and robustness of a pedestrian recognition result in a camera-cross environment.

Provided is a quick detecting method for moving objects in a dynamic scene. The quick detecting method for the moving objects in the dynamic scene comprises carrying out sequence interframe registration on moving images by utilizing CenSurE feature points and a homography transformation model, obtaining a registering frame of a former frame taking a current frame as reference, carrying out subtraction on the registering frame with the current frame to obtain a frame difference image to generate a foreground mask, building a dynamic background updated in real time according to space distribution information of the foreground mask in the current frame, obtaining a background subtraction image based on a background subtraction method, carrying out statistics on the probability density of the gray level of each pixel in the frame difference image, when the sum of the probability density of the gray level of a pixel is larger than 2phi(k)-1, taking the gray level as a self-adaptation threshold value, judging pixels with values of gray levels larger than the threshold value as foreground pixels, and otherwise judging the pixels as background pixels. The quick detecting method for the moving objects in the dynamic scene can reach the processing speed of 15frame / s and can obtain relatively integral moving objects under the premise that the detecting speed is ensured, and therefore, index requirements such as rapidity, noise immunity, illumination adaptation, target integrity and the like of the detection of the moving objects in the dynamic scene can be met.

Techniques for determining similarity between error reports received by an error reporting service. An error report may be compared to other previously-received error reports to determine similarity and facilitate diagnosing and resolving an error that generated the error report. In some implementations, the similarity may be determined by comparing frames included in a callstack of an error report to frames included in callstacks in other error reports to determine an edit distance between the callstacks, which may be based on the number and type of frame differences between callstacks. Each type of change may be weighted differently when determining the edit distance. Additionally or alternatively, the comparison may be performed by comparing a type of error, process names, and / or exception codes for the errors contained in the error reports. The similarity may be expressed as a probability that two error reports were generated as a result of a same error.

The invention discloses a tumble detection method based on human body posture state judgment. The tumble detection method comprises the following steps: extracting a target region of an image after binarization processing, extracting Zernike moment features, the height, the width, the aspect ratio, the target region area, the target region perimeter, the outline area, the outline perimeter, a frame difference maximum row and column characteristic variable and a moving object height feature variable of an object from the target region, respectively sending the extracted variables to two offline trained classifiers for judging moving conditions of a monitored person, and finally judging the real moving condition of the monitored person. According to the tumble detection method, an image processing technology is completely used to accurately distinguish posture states of walk, crouch and tumble, and the requirement of realtime processing on a hardware platform with lower performance can be met.

The invention discloses a flicker detection method and a device thereof and the device can carry out real-time detection without affecting the image quality. The method comprises the following steps of: reading a current frame, and obtaining the vertical waveform of the current frame; carrying out frequency spectrum analysis on the vertical waveform or carrying out frequency spectrum analysis on the difference (frame difference waveform) of the vertical waveform of the current frame and the vertical waveform of the historical frame; if the peak exists on the frequency spectrum, judging whetherthe flicker exists or not according to the approaching degree of the frequency of the peak corresponding point and the theoretical peak frequency point.

A digital acoustic signal coding apparatus, a method of coding the digital acoustic signal, and a recording medium for recording a program of coding the digital acoustic signal are respectively realized. It is possible to provide the digital acoustic signal coding method and apparatus, in which, corresponding to the difference between the sampling frequencies of the input acoustic signal, short blocks can be suitably classified into groups without deteriorating sound quality and the suitability of using either long / short blocks can be judged. The coding apparatus is composed of a calculation medium for calculating the sensation entropy of an input acoustic signal per each of the respective short sensation blocks; a sensation entropy sum total calculation medium for obtaining a total sum in a frame of the sensation entropy; a comparison medium for comparing an absolute value of the difference between the respective total sums of the sensation entropy of successive two frames with a previously determined threshold value; and a long / short block judgment medium for judging whether a long block or short blocks should be used to convert a block of the input acoustic signal on the basis of the comparison result.

The invention is directed to a method, apparatus, system and computer program product for analysing video images of a sports motion and, in particular, to identification of key motion positions within a video and automatic extraction of segments of a video containing a sports motion. Video data displaying a sports motion comprises data representative of a number of image frames. For a number of image frames, one or more frame difference measures are calculated between the image frame and another image frame. The frame difference measures are analysed to identify a plurality of image frames that each show a key position of the sports motion. The segment of the video containing the sports motion showing the sports motion is between two of the image frames showing the key positions. Frame difference measures may be calculated based on pixel difference measures or optical flow techniques.

The invention provides a method for detecting group crowd abnormal behaviors in video monitoring. The method comprises the steps of: video target detection: obtaining video objects through edge information difference detection in successive frames, and obtaining video objects with movement change through frame difference of a foreground frame and a background frame, obtaining a relatively accurate movement target by combining two video object detection results; video target tracking: tracking targets to obtain corresponding movement tracks through a video particle-based long-period movement estimation method; group crowd detection: carrying out spectral clustering analysis on the distance between the tracks and advancing speed information through movement characteristics of the group crowd in video; and identification of group crowd abnormal behaviors: establishing a model for crowd tracks by using an MGHMM (Mixed Gaussian Hidden Markov Model), and identifying blockage and fall through sudden change of a normal track. The invention integrates technologies of crowd target detection, group target track, mode identification and machine learning.

The invention discloses a meter pointer angle identification method based on image processing, belonging to the field of image processing and solving the problem of large frame difference errors when a meter pointer is detected by adopting an absolute angle method. The meter pointer corner identification method comprises the following steps of: 1, acquiring a gray level image of a meter as an original image; 2, preprocessing to obtain a binarization image; 3, obtaining a preliminarily extracted meter panel scale image; 4, carrying out Hough circle transformation and determining a circle center of a meter panel; 5, according to the circle center of the meter panel, carrying out polar coordinate transformation on the original image to obtain a polar coordinate system image; 6, analyzing an angle value scale of the y axis of a polar coordinate system and establishing a scale angle sequence; 7, searching the position of a pointer in the polar coordinate system to obtain a pointer binarization image; and 8, obtaining an angle value of the pointer binarization image towards the y axis under the polar coordinate system according to the scale angle sequence obtained in the step 6, outputting actual data of pointing of the pointer of the meter panel, and finally identifying the angle of the meter pointer.

A system and method for effectively encoding and decoding electronic information includes an encoding system with a tiling module that initially divides source image data into data tiles. A frame differencing module then outputs only altered data tiles to various processing modules that convert the altered data tiles into corresponding tile components. A quantizer performs a compression procedure upon the tile components to generate compressed data according to an adjustable quantization parameter. An adaptive entropy selector then selects one of a plurality of available entropy encoders to most effectively perform an entropy encoding procedure to thereby produce encoded data. The entropy encoder may also utilize a feedback loop to adjust the quantization parameter in light of current transmission bandwidth characteristics.

The invention discloses a video stream-based license plate recognition method, comprising the following steps: S1, detecting triggering, namely defining a rectangular region of interest, obtaining a binary image in the rectangular region of interest by a frame difference, and judging whether a vehicle passes through the region according to the result of the binary image; S2, license plate location, namely carrying out license plate detection in the detection region in the image, and cutting a detected license plate out to obtain a license plate image; S3, license plate pretreatment, namely carrying out license plate inclination correction and accurate license plate cutting on the license plate image; S4, license plate character cutting, namely cutting the character on the license plate image, and decomposing the character into a single character image; S5, license plate character recognition; S6, repeating operations from the steps S2 to S5 on each frame of image to obtain a plurality of license plate recognition results; and S7, result voting. By the video stream-based license plate recognition method, license plate recognition is carried out on the basis of a plurality of frames of images. Thus, the risk caused by a recognition error of a single image is greatly reduced, and the overall recognition rate of the license plate is improved.

The invention discloses an automatic extractor hood speed regulating control method based on image analysis and an automatic extractor hood speed regulating control system based on mage analysis. The control method specifically comprises the following steps: (1), by virtue of an image acquisition unit, collecting an oil fume image of the upper part of a gas stove at a fixed frequency f; (2), by virtue of an image processing unit, extracting edge information by taking frame difference image of the collected oil fume image and an oil fume-free background image as a signal source; (3), by virtue of a main controller, performing quantitative analysis by taking the edge information transmitted by the image processing unit as an image feature value, thereby obtaining a frequency signal required by a frequency conversion fan; and (4), by virtue of the frequency conversion fan, regulating the rotation speed according to the received frequency signal sent by the main controller. By virtue of an image analysis means, the concentration of oil fume generated by the gas stove is detected in real time, the rotation speed of the fan is regulated and closed-loop automatic control is realized.

The invention discloses a video key frame extraction method based on color quantization and clusters. The method comprises the steps of loading video data flow; conducting single frame scanning on video flow; conducting the color quantization on obtained frame images, and extracting main color features of the frame images going through quantization; calculating similarity of adjacent frames so as to obtain adjacent frame difference; conducting shot boundary detection according to the adjacent frame difference; conducting shot classification on intersected shots and extracting a representative frame of each shot; and conducting compression clustering on the sequence of the representative frames so as to obtain a key frame sequence. According to the method, the color quantization is conducted on the single frame images so that main color of the images is extracted, frame difference calculation is conducted through the cluster feature similarity calculation method based on color features of the clusters so that the shot boundary detection is realized, and finally clustering according to the compression ratio is conducted on the extracted representative frames. Due to the fact that he whole process is low in dependency on video formats and types, the method has good universality and adaptability, is simple in calculation and low in space consumption, and can effectively avoid the phenomenon of key frame selection redundancy, control the number and quality of the key frames, and realize control of the video compression ratio.

The present invention discloses a parking stall detection method. The method comprises: collecting a driving-in image when there is a vehicle driving in a detection area; after the remaining time of the vehicle exceeds a certain time, collecting a remaining image when it is continuously detected that the remaining time of the identified vehicle feature appears at the same position or in a template matching frame exceeds a certain time; collecting a driving-out image when the displacement distance of the identified vehicle features exceeds a preset distance threshold value or the area of the frame difference at the detection area exceeds a preset area threshold value, and determining whether the vehicle features are close to the position in the remaining image or not; and setting the parking stall as an unoccupied parking stall or updating the driving-out image according to the current collected image of the detection area. Therefore, the parking stall state detection efficiency is improved on the premise of reducing the maintenance quantity of a device.

The invention discloses an infrared image-based weak and small moving target detecting method. The method disclosed by the invention comprises the steps of median background difference image generation, accumulated difference image generation, difference image segmentation, periodical motion area elimination and false target elimination. Concretely, the method comprises: establishing a grayscale histogram for each pixel to count the frequency of the appearance of the pixel on each grayscale in the latest time period, calculating a median to obtain median background, and calculating the absolute frame difference between the median background and a detected frame to obtain a median background difference image; calculating the frame difference between each two adjacent frames continuously, and accumulatively adding difference images to obtain an accumulated difference image; eliminating interference from periodical motion areas, and highlighting a real moving target; and finding a reasonable target track by using motion and grayscale consistency of the target. The method disclosed by the invention can accurately and continuously detect the weak and small moving target day and night at a remote distance, has high real-time performance and high robustness, and is easy to promote in application field of military robot battle reconnaissance, video monitoring and the like.

The invention discloses a system for detecting a moving object, which comprises a distance map acquisition module, a total map enhancement module and a detection module; wherein the distance map acquisition module is used to carry out inter-frame differentiation to the input video image data and to build up a background module according to the frame differentiated image to obtain a background image; the current image and the background image are compared to obtain a distance map; the total map enhancement module is used to increase the disparity of the pixel values of the pixel point of the moving object and the noise pixel point on the distance map to obtain a totally enhanced distance map; and the detection module is used to determine the enclosing rectangle of the moving object in thecurrent image according to the totally enhanced distance map. The system improves the stability of the background module and enables the background module to be free from the influence of the moving object at the current scene, and the system is suitable for the detection of moving objects at very low contrast.

The invention discloses a deep learning-based intelligent indoor intrusion detection method and system. The method comprises the following steps of establishing a BP neural network model; obtaining difference images between adjacent frames in a monitoring video picture by utilizing a frame difference algorithm; carrying out binarization processing on the obtained difference images and extracting change foreground area images, under a static background, in the processed images; detecting and identifying whether human shapes exist in the extracted change foreground images or not; when the human shapes exist, detecting and extracting face area images from the change foreground areas; identifying whether the extracted face area images are user images or not; and when the face area images are not the user images, determining the face area images as non-user intrusion and sending alarm signals to the users. The method and system are strong in ability of resisting the interference of other moving objects and low in mis-judgement rate, and can be used for carrying out massive video data analysis and correctly carrying out intrusion detection and identification.

An intelligent monitoring and alarm method based on moving object detection has the following steps: step 1, picking up the grey scale information of a video signal; step 2, initializing a background matrix; step 3, carrying out background-based object detection; step 4, updating the background. The invention detects an initial sequence through a frame difference method and establishes the initial background according to the detection result; after the initial background is established, a background method is used to detect a moving object; the background model adopts a mean-based difference comparison method instead of a probability threshold value comparison method; the background updating is carried out in two parts, namely a foreground point part and a background point part. The method capable of automatically updating the background can detect the moving object in time with effectiveness and timeliness, thereby avoiding a 'ghost shadow' to a great extent. Furthermore, the invention can meet the requirements of multi-path real-time monitoring with simple operation and high speed.

The invention provides a TV station logo training method and an identification method. The TV station logo training method comprises the following steps of: (1) determining the external rectangle of a station logo in a training video according to a frame difference method; (2) carrying out edge detection on the external rectangle of the station logo to obtain edge points of the station logo; and (3) calculating distance transformation matrix Df of the station logo according to the edge points of the station log. According to characteristics obtained by the TV station logo training method, the position of the station logo in the video is determined only by the single-frame video information of the video to be detected not by continuous multiframe video information in the identification process; moreover, under the conditions that the shape and the size of the station logo have subtle changes, the station logo still can be successfully identified.

The invention discloses an adaptive noise intensity video denoising method which is based on motion detection and is embedded in an encoder. The method comprises the following steps: (1) taking a sum of regularization frame differences in a neighborhood as an observed value, dividing input pixels into a static pixel and a dynamic pixel and using filters in different supporting domains for the two kinds of the pixels, wherein a filtering coefficient is adaptively determined according to noise intensity and an image local characteristic; (2) taking a single DCT coefficient or the sum of the several DCT coefficients as the characteristic, using AdaBoost as a tool to construct a cascade-form classifier and using the classifier to select a static block; (3) establishing a function model of connection between DCT coefficient distribution parameters of the video noise intensity and the static block and using the model to estimate the noise signal standard difference. By using noise intensity estimation embedded in the video encoder and a noise reduction technology provided in the invention, few computation costs can be used to acquire the parameters and the information needed by noise filtering. A time efficiency is good. Because a reliable clue is used to determine whether the pixels accord with a static hypothesis, the filter of the invention can effectively filter the noise and simultaneously maintain marginal sharpness of the static image. And motion blur caused by filtering in a motion area can be avoided.

The invention relates to a stereoscopic vision based acquisition method of the congestion degree of bus passenger flow, comprising the following steps of: (1) obtaining a left synchronous video image and a right synchronous video image by a left parallel synchronous camera and a right parallel synchronous camera, and processing left and right synchronous video image sequences by a synchronous video image collection module; (2) computing a parallax image of a passenger flow scene by a parallax computation module according to the left and right synchronous video image sequences, and prestoring a background parallax image; (3) judging the sort of a current frame by a foreground object generation module according to the frame difference motion characteristic of the passenger flow scene, dynamically updating the parallax image of a background frame, and computing the parallax image of a foreground object in the scene according to a foreground frame; (4) carrying out position mapping for the foreground object by a oreground object mapping module so as to obtain a ground plane mapping image of the foreground object; (5) carrying out object characteristic statistics for the ground plane mapping image by a congestion degree estimation module, obtaining a congestion degree estimation value according to an object area and an edge characteristic; and repeatedly and circularly executing the steps from step (1) to the step (5) to complete the real-time collection of the congestion degree of the passenger flow.

The invention provides a regulating circuit and a regulating method capable of reducing power consumption of a screen. The regulating method comprises the steps of reading pixel values of current frame and previous frame from a frame storage unit by a difference calculation unit, calculating frame difference information, and sending to a frame difference storage unit; storing M pieces of frame difference information and sending the stored frame difference information of the current frame and the previous (N-1)th frame to a display scene judgment unit by the frame difference storage unit, and judging current display scene according to the frame difference information of the current frame and the previous (N-1)th frame and the configuration information and sending a judgment result to a display frame rate control unit and a display control unit frequency controller by the display scene judgment unit; and regulating the display frame rate by the display frame rate control unit according to the configuration information and the judgment result to change the refreshing rate, and regulating the working frequency of the display control unit by the display control unit frequency controller. According to the invention, the refreshing rate of the display screen and the working frequency of a display control circuit can be dynamically regulated according to different display scenes.

The invention discloses a mobile platform ground movement object detection method including the steps as follows: (1) respectively calculating SWIFT characteristic description of a (n-m)th frame image, an nth frame image and a (n+m)th frame image of an image sequence; (2) by taking the nth frame image as a reference, respectively registering the (n-m)th frame image and the (n+m)th frame image to obtain registration images; (3) respectively conducting space multiscale movement significance tests on the two registration images with the nth frame image; (4) calculating a difference absolute value image of the nth frame image and the (n-m)th frame registration image, and the difference absolute value image of the nth frame image and the (n+m)th frame registration image; (5) respectively calculating a difference image of the (n-m)th frame and the nth frame as well as a difference image of the nth frame and the (n+m)th frame; (6) mixing the two difference images to obtain a three-frame difference image and then conducting binaryzation; and (7) conducting morphological processing on a binary segmented image and then conducting region marking so as to obtain a final detection result image. The mobile platform ground movement object detection method solves the background compensation problem under the condition of a moving platform and has very good robustness on the light change and background interference in a scene as well as image deformation caused by platform movement.

The invention provides a natural background video sectional drawing method. The method comprises the steps that 1) the method automatically selects a key frame, specifically the method automatically selects the key frame in a video by calculating inter-frame difference; 2) a key frame Trimap map is generated, specifically a user us allowed to provide useful information through a small amount of manual interaction and a GrabCut segmentation algorithm and a Gaussian blur method are used to semi-automatically generate the Trimap map on the key frame; 3) a middle frame Trimap map is generated, specifically bidirectional optical flow propagation is used to generate the Trimap map of each middle frame, and color, gradient and other information are used to correct errors generated by optical flow propagation; and 4) according to the Trimap maps of all video frames and a Bayesian sectional drawing algorithm with added three-dimensional space-time optimization, the alpha mask map of each frame of the video is calculated, and the alpha mask map and a new background are synthesized to acquire a new synthesized video. According to the invention, the method has the advantages of fast speed and less user interaction, and the finally acquired new synthesized video has the advantages of good time coherence, high reality and the like.