33 results about "Video stitching" patented technology

1. Name of the product in this design: Visible light splicing monitoring equipment. 2. Purpose of this design: To use multi-channel video stitching for wide-angle observation of a scene. 3. The key design features of this product are its overall shape. 4. The image or photograph that best illustrates the design's key points: a 3D model.

A video coding mechanism for view-dependent video coding is disclosed. The mechanism includes receiving a spherical video signal spliced from a multi-directional video stream. The spherical video signal is mapped into a plurality of sub-picture video signals, each of which includes a sequence of sub-pictures. The plurality of sub-picture video signals are coded into a plurality of sub-picture bitstreams such that, when decoded in a decoder, each pixel in each sub-picture has a value equal to a corresponding pixel in a whole picture composed of the sub-pictures after decoding. The plurality of sub-picture bitstreams form a plurality of sub-picture tracks, and it is indicated that the sub-picture tracks conform to a specific video profile. The sub-picture bitstreams are sent to a decoder to support decoding and displaying a virtual reality video view.

This application discloses a panoramic stitching method, apparatus, device, and storage medium based on multi-camera systems, relating to the field of image processing technology. The method includes: acquiring calibration images collected by multiple cameras, wherein the calibration images include common calibration areas of adjacent cameras; identifying calibration board corner features in the calibration images and obtaining calibration parameters for each camera based on these features; generating a configuration file according to the calibration parameters; and processing the video streams collected by the multiple cameras based on the configuration file to obtain a panoramic video stream. This method solves the problems of low calibration accuracy and reliance on manual experience and complex online calculations in existing technologies under long-distance, large field-of-view conditions. It achieves standardization and automation of the calibration process and enables efficient, high-precision real-time panoramic video stitching on embedded devices, reducing the system's dependence on high-performance processing platforms and overall cost.

The embodiment of the invention provides a bucket wheel machine unattended system and method based on video stitching, and the system obtains the global high-precision visual information of a bucket wheel machine operation area and an accurate material pile semantic segmentation mask image through the panoramic image stitching and image recognition of the original video streams of a plurality of cameras. Meanwhile, the external parameters of the dynamic camera are solved in real time by combining with the PLC pose data of the bucket wheel machine, the two-dimensional semantic segmentation mask graph and the synchronized laser radar point cloud are efficiently fused, the material pile edge features are accurately mapped to the three-dimensional space, the high-precision three-dimensional material pile edge point cloud is obtained, and therefore the defects of the laser radar in semantic recognition are overcome. And finally, based on the three-dimensional material pile edge point cloud updated in real time and the current pose data of the bucket wheel machine, the system can intelligently carry out operation path planning, the operation path planning is decomposed into PLC control instructions, and autonomous, safe and accurate operation of the bucket wheel machine in the complex and changeable material pile environment is achieved.

A kind of distributed video splicing device, it is related to video splicing technical field, including with several local input modules of external image device connection, with the management module of several local input modules connection, with the output module of management module connection, with the network streaming media input module of management module connection, output module is connected with external display device, local input module includes with the input interface of external image device connection, with the input conversion chip of input interface electrical connection, with the input FPGA of input conversion chip electrical connection, input FPGA is electrically connected with management module, input conversion chip is used to output RGB signal to input FPGA.The technical scheme provided by the utility model aims to realize integrated design, and can solve the technical problem of network streaming media data processing.

The invention discloses a time-space consistency video stitching method and device, and relates to the technical field of video stitching. The method comprises the step of realizing high-precision video stitching through dual-channel feature extraction and spatio-temporal joint optimization. The method comprises the following steps: firstly, respectively extracting shallow layer and deep layer features of two paths of video sequences by adopting a StarNet-s050 model; constructing a space distortion module comprising a context correlation layer, RegressNet1 and RegressNet2 to generate a space distortion matrix before optimization, and obtaining a final space distortion matrix through an auxiliary training head; a time sequence distortion module is used for extracting space-time dynamic features, and space-time smooth increment is calculated in combination with smooth constraints; and finally, inputting the corrected space-time distortion matrix and an original video into a splicing module to generate a fusion sequence. According to the method, the problems of deformation difference of a multi-view video in a space dimension and fuzziness on a time axis are effectively solved, and the video stitching quality is remarkably improved through hierarchical feature fusion and space-time consistency modeling.

According to embodiments of the present disclosure, methods and apparatuses for remote assistance are provided. The method includes presenting an assistance interface associated with an autonomous vehicle; and presenting, in the assistance interface, a multi-path video stream corresponding to a plurality of viewing perspectives of the autonomous vehicle based on video data received from the autonomous vehicle, wherein the multi-path video stream includes a first video stream corresponding to a forward viewing perspective and a second video stream corresponding to a rearward viewing perspective, the first video stream and the second video stream are generated based on a video stitching model associated with the autonomous vehicle, the video stitching model is constructed based on a plurality of images captured by a plurality of cameras deployed on the autonomous vehicle. In this way, embodiments of the present disclosure can improve the reliability of remote assistance, thereby improving the driving safety of the autonomous vehicle.

The invention relates to the technical field of computer vision, and discloses a multi-camera video stitching method based on regionalized symmetric weighting. The method comprises the following steps: realizing multi-camera frame-level synchronous acquisition (time sequence jitter is less than or equal to 10ms) through an event-driven multi-thread mechanism; obtaining distortion parameters (k1, k2 and k3) by adopting a Zhang Zhengyou calibration method, and correcting the image in real time according to a nonlinear model; dynamically constructing a fusion region: firstly expanding pixels of an overlapping region We belonging to [100, 200] to compensate a registration error, then generating an optimal suture line through a graph cutting method, forming a suture line region through morphological expansion, and finally dividing the optimal suture line region into a suture line region and a transition region; during partition fusion, an image source with the minimum pixel difference is directly selected in a suture line area to eliminate ghosts, and a dynamic boundary weight strategy is adopted in a transition area to achieve seamless transition. Finally, two fusion problems of ghosting elimination and splicing seam elimination are realized based on the two, image transition becomes smoother and more natural, and the method is suitable for panoramic security and protection monitoring and the like in industrial complex scenes such as wind power and the like.

The application discloses a kind of multi-view video splicing methods based on deep learning, steps are as follows: first, using Airsim simulator is collected at the image and depth data of set virtual public viewpoint, generates the dataset for video splicing task, and the image is cylindrical projection and so on Preprocessing.Then, using convolutional neural network designs artifact elimination module and smooth transition module respectively, the former considers the feature correlation of overlapping area, and the overlapping area is aligned by viewpoint regression to eliminate the artifact after fusion;The latter according to the feature information of image, the deformation rule obtained in the overlapping area is propagated to the non-overlapping area, to guide the smooth transition between regions to improve visual perception.Finally, according to the predicted displacement field, the original viewpoint image is distorted and transformed, and the splicing result is obtained by weighted linear fusion.The application can achieve real-time splicing performance while removing splicing artifacts, and can meet the online splicing needs in practical applications.

The invention discloses a parallel video stitching method based on multi-band fusion, which comprises the following steps of: reading a plurality of video files in parallel, aligning timestamps and unifying the sizes of image frames to obtain an image frame sequence corresponding to the plurality of video files, and storing the image frame sequence into a cache pool; reading each image frame sequence, judging whether the obtained image frame is an initial frame or not, if so, extracting image features for feature matching, estimating camera parameters and taking as a registration template, and otherwise, entering the next step; the processing thread multiplexes the template to process subsequent image frames in parallel, and a splicing result of the image frames at the same moment in the multiple video files is obtained; splicing results of all the image frames are synthesized according to the time sequence, and a final video is obtained. According to the method, the real-time requirement of the video can be met while the video splicing effect is ensured.

A video stitching system combines video from different cameras to form a panoramic video that, in various embodiments, is temporally stable and tolerant to strong parallax. In an embodiment, the system provides a smooth spatial interpolation that can be used to connect the input video images. In an embodiment, the system applies an interpolation layer to slices of the overlapping video sources, and the network learns a dense flow field to smoothly align the input videos with spatial interpolation. Various embodiments are applicable to areas such as virtual reality, immersive telepresence, autonomous driving, and video surveillance.

The invention relates to a multi-dimensional video content intelligent editing method and device, equipment and a medium. The method comprises the following steps: extracting multi-dimensional video features including visual features, audio features, semantic features and motion trail features according to a multi-camera video stream; fusing the multi-dimensional video features to obtain a fused feature vector; classifying the video frames based on the fusion feature vector to generate a scene label sequence; reasoning and generating an editing decision sequence according to the scene label sequence through a decision tree reasoning engine; positioning a lens switching point according to the editing decision sequence to obtain an editing time point set; and splicing the multi-camera video stream based on the set to generate a target edited video. By adopting the method, the problems of one-sided editing decision and poor scene adaptability in the prior art can be solved, the content integrity and the scene adaptability of multi-camera video editing are improved, and the high-quality presentation requirement is met.

This application discloses a video stitching method, apparatus, electronic device, and storage medium for multi-camera surveillance, aiming to solve the problems of poor regional adaptability, inaccurate feature matching, lack of multi-dimensional synchronization, single parameter tuning mechanism, and insufficient balance between data storage and security in the prior art, and to achieve high-precision, seamless, and adaptive panoramic stitching of large-scale multi-camera surveillance videos.

The present disclosure relates to the technical field of image processing, and provides a video splicing method, device, equipment, medium and program product. The video splicing method comprises: acquiring source video frames; performing correction preprocessing on each source video frame to obtain a corrected video frame and a corrected video frame group; sequentially acquiring an overlapping area of a first corrected video frame and a second corrected video frame of each corrected video frame group; acquiring an optical flow of the overlapping area according to a deep learning model; sequentially acquiring a non-overlapping area of the first corrected video frame and the second corrected video frame of each corrected video frame group; completing the optical flow of the non-overlapping area to obtain an overall optical flow; sequentially mapping the first corrected video frame in each corrected video frame group according to the overall optical flow to obtain a mapped video frame; and splicing the first corrected video frame and the mapped video frame in each corrected video frame group to obtain a spliced video frame. The video splicing method can improve the image splicing quality.

The invention discloses an embedded real-time video stitching method based on dynamic suture line optimization and time domain smoothing. The method comprises the following steps: performing multi-thread synchronous acquisition and preprocessing on left and right video streams; extracting features on the key frame and obtaining an initial homography matrix, smoothing the time domain of the homography matrix, and multiplexing or restricting updating when the matching quality is insufficient to suppress picture jitter; constructing an energy diagram in the overlapped region ROI (region of interest), searching an optimal vertical suture line with the minimum accumulated cost from top to bottom by utilizing dynamic programming, and reducing dynamic ghosting; a hard mask and a soft mask are generated in a suture line neighborhood and are respectively used for a high-frequency layer and a low-frequency layer for multi-band fusion, high-gradient side forced retention is performed on a position where a pixel difference exceeds a threshold value in a mixed band, and finally, a Laplacian pyramid is only constructed near an overlapping region, and memory copying is performed in a non-overlapping region. The real-time performance and the splicing visual quality can be considered.

The invention discloses a video stitching method and device based on optical flow interpolation. The method comprises the following steps: projecting a received multi-view video frame to a public cylindrical coordinate system to obtain a projection image under each view; for overlapped areas of the projection images under two adjacent visual angles, calculating bidirectional optical flow by using the optical flow estimation sub-network and then performing zooming, and performing optical flow interpolation on the two overlapped areas based on the zoomed optical flow to obtain a plurality of slice distorted frames; generating two refined optical flows and a visibility graph by using an optical flow refining sub-network based on the slice distorted frame and the scaled optical flow; calculating a fusion interpolation slice based on the refined optical flow and the visibility graph to generate a fusion image and cover the overlapping region to obtain a spliced video frame; and designing a composite loss function based on unsupervised learning to carry out network training. According to the method, ghosting and fracture artifacts at the video splicing boundary can be effectively eliminated, a high-quality panoramic video with time-space coherence is obtained, and the method is particularly suitable for wide-baseline camera deployment scenes such as automatic driving and security monitoring.

The application discloses a kind of camera equipment parameter-independent fast multi-view fish-eye image and video splicing method, comprising the following steps: S1, the fish-eye image of fish-eye camera shooting is preprocessed, the pre-processing includes image correction and inter-lens parallax optimization;S2, the image after pre-processing is registered using the method based on grid transformation;S3, image registration is optimized;S4, dynamic video splicing is carried out, and panoramic image and video are obtained.The application has lower computational complexity, can quickly capture the scene information around camera, can effectively improve the accuracy, adaptability and computational efficiency of image and video splicing, is conducive to being combined with computer vision and computer graphics and other computer tasks;The application method has wide application range and strong robustness, can still realize high-precision panoramic splicing quality in different types of natural scenes, and can be applied to related computer applications with high requirements for panoramic image and video quality.

The invention belongs to the technical field of large model application, and particularly relates to an intelligent video propaganda product design method based on a large model and a knowledge base. Aiming at the problems that the traditional video propaganda product design is long in period and low in standardization, and the existing large model video generation has inter-frame motion disorder, content illusion, weak demand analysis and the like, the invention provides a video propaganda product intelligent design method based on a large model and a knowledge base. The method comprises the following steps: preprocessing a natural language demand of a user to generate a standardized structured data set; generating a video script through a large script generation model, and segmenting the video script into a split script with a time sequence anchor point; generating an inter-frame motion optical flow field through an optical flow prediction model, and constraining a video large model to generate a coherent sub-mirror frame; and carrying out exception verification after video stitching, carrying out reverse matching on an exceptional fragment, regenerating and replacing a split script, and outputting a finished product until the verification is passed. According to the invention, video production efficiency and demand matching degree are improved, and picture fluency and real and compliant content are guaranteed.

The invention provides a panoramic video stitching method based on a multi-view camera. The panoramic video stitching method comprises the following steps: step 1, data preprocessing and feature extraction; step 2, triggering self-calibration; step 3, layered calibration: adopting layered calibration logic of first internal reference and second external reference, and reducing variable dimensions after decoupling; and 4, verifying and applying a calibration result. According to the method, the iteration convergence period can be greatly compressed, compared with the iteration redundancy problem caused by the fixed step length of a traditional L-M algorithm, the external parameter optimization speed is increased, the calculation power loss of a calculation chip is reduced, and the application requirements of middle-end and low-end vehicle-mounted hardware are met.

The invention discloses a video stitching method and device, equipment and a medium, and relates to the technical field of image processing, and the method comprises the steps: obtaining a first overlapping region and a second overlapping region; the first overlapping region is a region, which is overlapped with a previous frame of the video frame to be fused, in a previous frame of the target video frame, and the second overlapping region is a region, which is overlapped with the video frame to be fused, in the target video frame; fusing the abutted seam of the first overlapping area and the abutted seam of the second overlapping area to obtain a target abutted seam; and splicing and fusing the target video frame and the to-be-fused video frame by using the target splicing seam.

The application provides a synchronization method and system of distributed video control, the synchronization method comprises the following steps: obtaining the input signal of the first distributed control unit connected with the input end, determining whether the first distributed control unit is a synchronization source unit or a following unit according to the input signal; if the first distributed control unit is the synchronization source unit, taking the clock signal of the first distributed control unit as the reference clock, generating the field synchronization signal, outputting the reference clock and the field synchronization signal to the second distributed control unit connected with the output end, and synchronizing the second distributed control unit according to the reference clock; if the first distributed control unit is the following unit, outputting the reference clock and the field synchronization signal forwarded by the first distributed control unit to the second distributed control unit connected with the output end to synchronize. The synchronization method and system of distributed video control provided by the application can realize expansion by adding a simple USB cascade line in the distributed control unit without a complex interaction protocol.

The invention discloses a ship-borne panoramic video stitching method based on multiple distributed cameras, and relates to the technical field of ship intelligent navigation and computer machine vision, a plurality of network cameras are respectively deployed on a bow, a stern, a left board and a right board of a ship body, multiple paths of video streams are collected in real time through an RTSP protocol, and decoding and frame synchronous caching are carried out; dark channel defogging, image sharpening, OTSU binaryzation, Canny edge detection and edge refining processing are sequentially carried out on each path of video stream. A distributed ordinary network camera architecture is adopted, deployment is performed on the bow, the stern, the left side and the right side of the ship body, the existing installation space and conventional network equipment of the ship are fully utilized, a special panoramic camera does not need to be customized, unified access of video streams is achieved through an RTSP protocol, software splicing is completed through an edge computing server, and the whole process is simplified. Hardware purchase and transformation cost is remarkably reduced, a distributed structure supports modular extension and maintenance, overall operation of the system is not affected by faults of a single camera, and the system has high cost performance and maintainability.

The invention provides a pinhole image splicing truth value acquisition and splicing method based on heterogeneous camera combination, and the method comprises the steps: symmetrically installing two pinhole cameras on a support, installing a fisheye camera between the two pinhole cameras, calibrating the three cameras based on a double-spherical camera model, and carrying out the calibration of the three cameras, the method comprises the following steps: obtaining internal parameters and external parameters of three cameras, simultaneously acquiring a left-eye pinhole view, a right-eye pinhole view and a middle-eye fisheye view of a target scene by using the three cameras, performing equidistant rectangular projection on the middle-eye fisheye view, and obtaining a target scene according to the internal parameters and the external parameters of the three cameras. And respectively projecting the left-eye pinhole view and the right-eye pinhole view to a view plane of equidistant rectangular projection to complete unification of a multi-camera coordinate system and obtain a splicing true value. Core advantages are formed by means of hardware structure design and a heterogeneous camera image registration algorithm, images which are free of splicing seams and consistent in time sequence can be stably obtained, and powerful support is provided for precision verification of follow-up panoramic video splicing and truth value reference of model training.

The invention provides a video stitching method, electronic equipment, a readable storage medium and a computer program product. The video stitching method disclosed by the invention comprises the following steps: in response to a received fusion instruction of a multi-source video, acquiring an event time period of a collection object of the multi-source video and recording start time and recording end time of each video clip in the multi-source video; determining a reference video from the multi-source video by adopting a preset reference video judgment rule; performing time gap detection according to the event time period and the recording starting time and the recording ending time of each video clip of the reference video to obtain a gap time period; matching the gap time period with the recording starting time and the recording ending time of each video clip of other videos except the reference video in the multi-source video, and determining the shearing time range of other videos corresponding to the gap time period; and carrying out video cutting on other videos based on the cutting time ranges of the other videos, and splicing the other videos with the reference video.

The invention provides a video stitching and synthesizing method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining input data; wherein the input data comprises an original video sequence, a corresponding user text prompt set and a parallel audio stream; inputting the input data into a pre-trained video stitching and synthesis model, and outputting a mask sequence of each element of each frame; wherein the elements are in one-to-one correspondence with prompts in the user text prompt set; the splicing and synthesis model is obtained through multi-modal feature fusion and time sequence consistency constraint joint training, and is used for generating a pixel-level mask and synthesizing an output video; and performing foreground extraction and fusion processing based on the mask sequence of each frame and each element to generate a target synthetic video. Through the method provided by the invention, the target synthetic video is generated, and end-to-end intelligent video stitching and synthesis driven by user intention are realized.

The invention discloses a train positioning and loading system based on the video stitching technology and a control method thereof.The train positioning and loading system comprises a train, a stock bin, a camera, a weighing level gage, an electric unloading gate and an electric control system, the stock bin is arranged above the train, the camera is arranged between the train and the stock bin, and the weighing level gage is arranged above the electric unloading gate; the weighing level gage and the electric discharging gate are arranged at the lower part of the stock bin; the camera, the weighing level gage and the electric discharging gate are electrically connected with the electric control system through control cables. The system has the beneficial effects that the complete image of the carriage is obtained by utilizing the electric control system through the visual identification and video stitching technology, the charging amount is monitored in real time and the electric discharging gate is automatically controlled to be started and stopped in combination with the weighing level gage, the fine control of the charging weight and volume, the positioning of the train carriage and the identification of the charging volume are realized, and finally manual work is replaced; the maximization of the loading material quantity is realized, the loading efficiency is greatly improved, and the production benefits of enterprises are increased.

1. The name of the design product: video splicing processor. 2. The use of the design product: for video, image processing and control. 3. The design points of the design product: in the shape of the product. 4. The picture or photo that best shows the design points: perspective view 1.

The invention provides a video stitching method and device for an array camera and a medium, and the method comprises the steps: obtaining a panoramic video and a plurality of sub-videos shot by the array camera, and carrying out the moving object detection of the panoramic video and the plurality of sub-videos, so as to determine an optimal moving object; acquiring position information of each frame of image in the sub-video in the panoramic video according to the feature information corresponding to the optimal moving object; performing perspective transformation and image cutting on each frame of image of the sub-video to obtain a transformed sub-image, and splicing multiple frames of sub-images according to position information corresponding to each frame of image to obtain a plurality of spliced video streams; and performing moving object identification in the adjacent video streams, determining position deviation information corresponding to the adjacent video streams according to the motion trails of the same moving object in the corresponding video streams, eliminating splicing dislocation corresponding to the adjacent video streams according to the position deviation information, and completing splicing of the plurality of video streams. Splicing dislocation can be effectively eliminated, and the splicing precision is improved.

The invention discloses a target identification method based on a monitoring video, and relates to the technical field of target identification. The method comprises the following steps: acquiring a plurality of regional monitoring videos, extracting a preliminary monitoring image key frame sequence through an optimized inter-frame difference method, and screening through a secondary clustering method to obtain an optimized monitoring image key frame sequence; carrying out feature matching on the optimized monitoring image key frame sequence by adopting a scale invariant feature conversion algorithm to obtain a monitoring image sequence with the same feature, and carrying out image stitching by adopting a video stitching method based on a random sampling consensus algorithm to obtain a stitched monitoring image sequence; fusing the spliced monitoring image sequence through a video fusion algorithm based on an adaptive threshold to obtain a fused monitoring image sequence; and identifying the optimized key frame sequence and the fused monitoring image sequence through a target identification algorithm to obtain a first confidence coefficient and a second confidence coefficient, and obtaining a corrected confidence coefficient through an adaptive weighted fusion method.

The invention provides a hundred-million-pixel unmanned aerial vehicle (UAV)-borne multi-scale camera system and method. The system comprises an acquisition module, a calculation module, an integration module and an adjustment module. The system measures the depth of a target scene in real time through an acquisition module; the calculation module calculates a depth error according to the scene depth and a preset reference depth (based on a minimum parallax threshold); the integration module carries out integration on the error to generate a smooth baseline adjustment amount to avoid adjustment abrupt change; the adjustment module dynamically changes the camera spacing according to the adjustment amount by using a piezoelectric actuator platform, and minimizes the parallax. In combination with hardware-level color calibration, the system significantly reduces the video stitching error rate, realizes real-time output of a high-definition hundred-million-pixel video, and enhances the robustness and imaging quality of the system.