24 results about "Video image" patented technology

The invention discloses a front-vehicle road condition display system. The front-vehicle road condition display system comprises a plurality of first vehicular units, second vehicular units and roadside units; the first vehicular units are used for collecting safety information of a first vehicle in real time through a vehicle terminal, and sending the safety information to the roadside units, wherein the safety information at least comprises a road condition video image in front of the first vehicle; the roadside units are used for receiving the safety information sent by the first vehicular units, and forwarding the safety information to the second vehicular units; the second vehicular units are used for receiving the safety information forwarded by the roadside units and sent by the first vehicular units, and at least displaying the road condition video image on a second vehicle. The invention further discloses a front-vehicle road condition display method. By adopting the front-vehicle road condition display system, road condition information of a front vehicle can be seen in a vehicle following process to improve the driving safety of the vehicles.

A method for enhancing detection of a moving object. The method includes comparing a video image to a historical sequence of images obtained from the same scene or stationary source. The method uses statistical sampling to implement change detection and thresholding operations which extract moving objects from the scene. The method provides for a high degree of sensitivity and noise rejection to motion detection, and is self-adapting to variations in sensor characteristics and operating environments.

A video picture management equipment includes a first display control module and a second display control module which are connected to each other. The first display control module displays first video data on first video window, and second video data on the first video window at the same time according to a requirement; and the second display control module displays the second video data on a second video window, and the first video data on the second video window at the same time according to a requirement. A method for video picture management is also provided. By the above, the problems about the simple display mode and the unstable display effect during performing the PIP in the prior art are solved.

The invention discloses a behavior prediction method and device based on deep reinforcement learning, wherein the method comprises the following steps of: extracting an action characteristic of humanskeleton information through a frame of a video image; encoding the action characteristics according to the human body mechanism information; screening the effective part of the human body behavior process through the deep reinforcement learning so as to predict the behavior of people. According to the method, a local image block is extracted at an important position of a human body, and an ordered arrangement of features of the image blocks is utilized, so that the structural information of the human body is effectively utilized in the prediction process, and the accuracy and the performanceof the behavior prediction are effectively improved.

A digital image is obtained from a camera in a smartphone and is processed by a processor on the smartphone to modify a part but not all of the pixels in a frame of the digital image, which is a digital video image, which is modified by image processing in accordance with a modification mask and reversible modification instructions, and is transmitted to a second smartphone with a display screen to display the modified image. Image processing data are transmitted to the second smartphone to enable restoring the original digital video image from the partially reversibly modified image and to be displayed in unmodified form on the display screen.

The invention discloses a face feature point rapid alignment method based on an AAM, and belongs to the technical field of mode identification. The face feature point rapid alignment method of the invention comprises the steps of performing face feature point calibrating on an input face image so as to use the input face image as a training sample of a contour model; performing contour model training on the training sample by using an AAM active contour model method to obtain the contour model; inputting a continuous multi-frame video image to be processed; if the current frame is a starting frame, performing AAM search matching on the whole image to obtain a face feature point alignment result of the current frame and contour model; and if the current frame is not the starting frame, the face feature point alignment result of the previous frame is regarded as the starting position, and AMM searching and matching is carried out on the current frame to obtain the face feature point alignment result of the current frame and the contour model. The face feature point rapid alignment method of the present invention is applied to face recognition processing, the iteration initial position of the current frame is determined by the alignment structure of the previous frame, so that the search range and the number of iterations are reduced and the time consumption in the face feature point alignment is greatly reduced.

The embodiment of the invention discloses a bidirectional inter-frame prediction method and device, relates to the technical field of video image coding and decoding, and improves the coding and decoding efficiency. The method comprises: acquiring indication information, wherein the indication information is used for indicating that second motion information is determined according to first motioninformation, the first motion information is motion information of a current image block in the first direction, and the second motion information is motion information of the current image block inthe second direction; acquiring first motion information; determining second motion information according to the first motion information; and determining a prediction pixel of the current image blockaccording to the first motion information and the second motion information.

The invention discloses a road-occupying operation monitoring method based on target detection and pedestrian tracking, and the method comprises the following steps: setting the position of a target monitoring region according to a video image obtained by a road monitoring camera, and carrying out the continuous image extraction of an input video stream according to a preset period, using the trained target detector to detect the lane occupying behavior; when the target detector detects that a road occupying behavior exists in the video image, starting to detect people appearing in the video image, and when people are detected, starting pedestrian tracking, and recording each person appearing in the video image and a movement track of each person; and analyzing the pedestrian trajectory data in the video stream, judging whether the pedestrian trajectory data accords with the road-occupying operation behavior characteristics or not, if so, triggering an alarm, and if not, ignoring the road-occupying behavior. According to the invention, by combining the moving track information of the person appearing in the monitoring scene, the road-occupying operation behaviors are automatically and accurately analyzed, researched and judged, temporary lane occupation and road-occupying operation events can be effectively distinguished, and the false alarm rate is reduced.

The invention discloses a dynamic video fitting system which is characterized by being provided with a model video image mechanism, a member fitting room, a fitting member information base, a video processor and a video player, wherein the model video image mechanism is used for storing model video image data; the member fitting room is used for storing names of fitting members; the fitting member information base is used for recording and storing the information of the fitting members; the video processor is used for calculating the information of the fitting members and the model video image data to generate fitting video data correlative to the information of the fitting members; and the video player is used for playing fitting video data images of the members by video. The dynamic video fitting system has the remarkable effects of exhibiting the dynamic video effect of a commodity for a consumer, changing body characteristics of models in the model video image data according to the condition of the consumer to accord with the body characteristics of the fitting members, adjusting a fitting video picture according to the condition of the consumer so as to enable the consumer to feel a true fitting effect, setting an independent fitting room and an independent information base for each family member and storing the body characteristic information of the family members.

The invention discloses a switcher capable of automatically processing news simulcasting time extension. The switcher includes a base, a switcher main body is mounted at the top of the base; a coolingfan is mounted on one side of the power panel; a touch display panel is mounted on one side of the switcher main body; one end of the touch display panel is provided with a signal switching port. Oneside of the signal switching port is provided with a signal decoding port and a video image graph acquisition matching identification port. The invention discloses a switcher capable of automaticallyprocessing news simulcasting time extension. A CCTV1 signal is distributed into one path to be decoded to generate a video image picture; the system uses a video image identification technology. Firstly, news simulcasting corner mark identification and news simulcasting end identification of a fixed area in a news simulcasting picture are collected and stored, broadcasting and ending of news simulcasting are judged according to appearing and disappearing of corner mark and end identification graphs, and therefore complete rebroadcasting of news simulcasting and smooth broadcasting of subsequent programs are guaranteed.

The invention discloses an automatic analysis and calculation device for tracking precision of a photoelectric turret. A display is connected with a calculation processing unit; an external controller is connected with the calculation processing unit through a USB, a serial interface or Bluetooth; the calculation processing unit comprises a video input interface module, a calculation processing module and a data storage module; a photoelectric turret video image input connecting wire is connected with a photoelectric turret video output port and the calculation processing unit, and the video input interface module accesses a photoelectric turret output video, uniformly converts the video into a digital image to be processed and transmits the digital image to the calculation processing module; and the external controller inputs the field angle and image resolution parameters of a tested photoelectric turret to the calculation processing module to control the operation and stop of the calculation processing module, and a result processed by the calculation processing module is sent to the data storage module to be stored. The device is wide in application range, can be set according to parameters of different photoelectric turrets, and is suitable for tracking precision testing of all photoelectric turrets at present.

The invention is applicable to the technical field of computer vision, and provides a target object tracking method, a terminal device and a medium. The method comprises the following steps: respectively obtaining view angle images of a target object under multiple view angles; initializing a corresponding target tracking model for each visual angle image; loading a video image sequence to be tracked; in the process of loading each image frame, selecting a reliability model from each target tracking model, wherein the reliability model is a target tracking model of which the reliability scoremeets a preset condition; and positioning a target object in the current image frame based on the reliability model. According to the invention, tracking monitoring of the target object can be realized based on the multi-view target tracking model, so that a good tracking effect can be achieved in a relatively long time period; in the process of loading each image frame, the reliability model is selected from each target tracking model, and the target object is positioned based on the reliability model, so that the tracking accuracy of the target object is improved.

The invention provides a video imaging correction method based on odd-even positions, relates to the technical field of CCD detector video imaging correction, and solves the problem that the difference between a detector and odd-even channels of a video processor is difficult to correct by adopting a conventional radiometric calibration method in the prior art. The invention provides a video imaging correction method based on odd-even positions, and the method comprises the following steps: performing radiometric calibration in two times after power-on, respectively setting driving control signals of a detector at odd-numbered counting positions and even-numbered counting positions, and then distinguishing odd-even channels according to odd-even indication signals of accompanying data of a video processor; calibrating the difference of odd-even channels of the detector and the difference of odd-even channels of the video processor, and performing the calibration respectively. According to the method, the parity channels of the video processor are identified and corrected in the imaging processor, extra identification signals do not need to be added for output, and only parallel data and an accompanying clock need to be output.

The invention discloses a historical video real-time linear tracking method, which relates to the field of video image processing, and comprises the following steps of: 1, carrying out initializing and loading a library; step 2, carrying out linear tracking processing on historical videos; 3, releasing resources and unloading a library; step 4, ending. In the step 2, human factors are introduced,single-frame rapid playing is carried out on the video according to needs, target object extraction is carried out, a corresponding mark is marked at the position where the target appears, and when the target turns or suddenly changes, tracking is carried out through moving the target mark at the turning point again. Through the implementation of the invention, the precision of the target trackingalgorithm in a complex scene is improved, the problem of target tracking errors or target tracking missing in a complex picture can be solved, the method is not influenced by error timestamp information, and the method is simple and easy to popularize and apply.

The invention discloses a calculation method for block mass acquisition based on a neighborhood box, and the method comprises the following steps: S1, carrying out the preprocessing of a video, and obtaining a video image; s2, segmenting the video image into M small areas, and determining a seed point corresponding to each small area; s3, in a preset grid range, calculating the distances between the pixels and all the seed points, determining the seed points of pixel classification according to the distances, and constructing a block mass; and S4, updating the positions of the seed points until the positions of the seed points are located in the central area of the block mass. According to the invention, the image dimension is reduced; the method is advantaged in that the image is made tomore easily participate in calculation, some abnormal pixel points can be further removed, interference of the abnormal pixel points on extraction of the integral image characteristics is prevented, the block mass technology is further applied to video processing, change of similar blocks between frames can be compared, a block mass track is formed, and motion characteristics can be conveniently extracted.