149results about How to "Improve stitching effect" patented technology

The invention provides a video play method and a video play device based on character interaction. The video play method comprises the following steps: step 1, carrying out face detection on a video image to be played so as to obtain a face region; step 2, carrying out pose estimation on the face region so as to obtain pose parameters; step 3, according to the pose parameters, searching a matched viewer face image as a matched image from a database; and step 4, replacing the face region by using the matched image so as to obtain a new video image. According to the invention, the character interaction can be carried out between the viewer and a played video, thereby enhancing the participation degree of the viewer and promoting the attraction of the video.

The present invention discloses a multi-image stitching method. The method comprises the following steps of: performing downsampling processing on each to-be-stitched image to obtain a corresponding downsampling image, and determining an overlapping relationship and a relative position relationship between the downsampling images; according to the overlapping relationship and the relative position relationship between the downsampling images, determining a feature point matching set between any two of the to-be-stitched images; and according to the feature point matching set between any two of the to-be-stitched images, performing calibration on a camera position of a current to-be-stitched image by means of a bundle adjustment method, then fitting to obtain an optimal orthogonal projection plane according to all the to-be-stitched images and corresponding calibration results, projecting each to-be-stitched image on the optimal orthogonal projection plane to obtain a corresponding projection result and performing fusion on the corresponding projection results. The multi-image stitching method disclosed by the present invention has the advantages of high stitching rate and high stitching quality.

The invention discloses a method for realizing subtitle overlay screenshot based on deep learning, and belongs to the field of media technology. The method comprises the following steps: selecting a video interval for subtitle overlay screenshot from a video; locating and cutting out subtitles on each frame of image in the video interval; segmenting all subtitles and extracting key frames in eachsubtitle; performing similarity calculation on the key frames, and performing comparative duplication elimination according to a calculation result to obtain final subtitles; and stitching a first frame of picture in the video interval with the final subtitles in sequence to obtain a subtitle overlay screenshot. The method is low in error rate, high in processing efficiency and high in automationdegree.

The invention discloses an infrared image splicing method based on coarse-to-fine registration, and the method comprises the steps: carrying out the coarse registration of an infrared image, carryingout the fine registration of the infrared image, improving the image registration precision, and improving the image splicing effect, wherein in the coarse registration process, feature points are extracted by utilizing every two adjacent images to construct a homography matrix, and each image can use an untransformed original image as an extraction object of the feature points in the coarse registration process, so that the extraction quantity of the feature points is ensured. The fine registration process is based on a fine registration sequence for sequential registration; deformation errors existing in the coarse registration process are adjusted, the feature point extraction process is obtained on the basis of the rule that the percentage of the area of the minimum circumscribed convex polygon covered by all feature point pairs consistent with the feature point pairs and the area of an original image is in direct proportion to the image registration accuracy, and the precision offine registration is guaranteed.

A sewing machine for sewing ultra-wide thick plate belt materials comprises a frame, an upper punching shear device and a lower punching shear device mounted on the frame, the upper punching shear device comprises an upper bottom plate, an upper mould clamp plate fixedly mounted on the upper bottom plate, an upper pressure plate connected with the upper mould clamp plate by an upper elastic device, and an upper punch fixedly mounted on the upper mould clamp plate, the top of the upper punch is arranged into an upper guide hole aranged on the upper pressure plate; the lower punching shear device comprises a lower bottom plate, a lower mould clamp plate fixedly mounted on the lower bottom plate, a lower pressure plate connected with the lower mould clamp plate by a lower elastic device, and a lower punch fixedly mounted on s the lower mould clamp plate; a punching shear hydraulic cylinder is arranged between the lower bottom plate and the frame, the upper bottom plate is connected by the lower bottom plate by a guide shaft, the punching shear cutting edge curves of the upper punch and the lower punch are approximately circular curves fit by multiple points. The sewing machine for sewing ultra-wide thick plate belt materials of the invention can be suitable for punching and sewing thin and narrow belt materials and ultra-wide thick plate belt materials.

The invention provides a panoramic image splicing method in an airport environment. The panoramic image splicing method comprises the following steps: firstly, preprocessing acquired images; adpotinga color difference correction method based on a YUV space to realize color consistency among multiple cameras; projecting a camera image to a cylindrical plane, and detecting feature points in the image by adopting a feature point detection algorithm for images after cylindrical projection; achieving rough matching of the feature point pairs in adjacent images; removing error feature point pairs through an RANSAC algorithm, calculating an accurate homography matrix between images to achieve uniform transition splicing between registered images through a gradual-in gradual-out fusion technology; carrying out canny edge detection on the panoramic image, selecting the longest line segment as the skyline position, and fitting a skyline smooth curve by using a least square method so that imagedistortion caused by splicing can be eliminated. A grouping iteration splicing method is adopted, error accumulation in the multi-path image splicing process is eliminated, and the splicing success rate and the splicing effect are improved.

The invention provides an optimal seaming line searching method for image splicing. A vector having the same dimension as the height of an image is defined to be a chromosome being an alternative seaming line by using a difference image at an image overlapped region as searching space, wherein each gene of the chromosome expresses an optimal splicing point at each row in the difference image. The fitness value of the chromosome being an alternative seaming line is defined to be a reciprocal of the sum of pixel values of locations of the splicing points on the difference image. Meanwhile, each chromosome is divided into a plurality of daughter chromosomes randomly; all daughter chromosomes are optimized respectively; and optimization results of the daughter chromosomes are combined to obtain an optimal seaming line. On the basis of the random grouping optimization strategy, large-scale decision variables can be processed and the seaming line quality can be improved effectively.

Disclosed are a method and system for calibrating an assembly error of a panorama camera. The method comprises the steps of establishing a monocular calibration model, establishing a binocular three-dimensional calibration model, and calculating the assembly error of the panorama camera based on the monocular calibration model and the binocular three-dimensional calibration model. Based on the calibration model of a standard camera, the angle between adjacent camera coordinate axises and the error of rotation and translation of all coordinate systems are calculated, and the assembly error of the panorama camera can be accurately and concisely described through the two parameter indexes. Furthermore, the invention provides a feasible method for solving the assembly error. The splicing accuracy problem caused by the camera assembly error is preferably solved in production and application of the panorama camera, therefore, the splicing effect is improved.

The invention relates to a preparation method of tubular porous bacterial cellulose. According to the preparation method, double layers of oxygen-permeating materials are nested to form a hollow oxygen-permeating mold for culturing bacterial cellulose, an inner pipe and an outer pipe of the mold can shrink/relax along the radial direction in a periodical alternation manner by accurately adjusting the pressure and the oxygen volume concentration of the internal pressure gas of the inner pipe and the external gas of the outer pipe of the hollow oxygen-permeating mold; and simultaneously sustained release microspheres which can release oxygen or a sugar source slowly are added into a culture solution so as to prepare the tubular porous bacterial cellulose. According to the preparation method, the effective area in the bacterial cellulose is expanded by more than one time, the culture period is shortened, the outer wall of the tubular bacterial cellulose is compact, the suturing performance of materials is improved, the probability of intimal hyperplasia and thrombus at the suturing position after the tubular porous bacterial cellulose is planted into a human body is lowered, and a formed porous structure is beneficial to the entering of cells and nutrient substances into the materials; and the material preparation period is short, the preparation process is environment-friendly, simple and fast, and the preparation cost is low.

The invention discloses an onboard 360-degree around-viewing display method and system. The method comprises the following steps of: S1, enabling a projection module to subject images acquired by respective cameras to viewpoint transformation by looking up a lookup table and to project the images onto a vehicle top view; S2, subjecting the four camera images to optical alignment by using an exposure gain algorithm; S3, splicing the splicing seams of the images acquired by the adjacent cameras by using a splicing algorithm to obtain the vehicle top spliced view. The method reuses the splicing seam of a last frame when the splicing seam at a certain position is not large in adjacent timing sequential difference so as to reduce an influence of noise on the calculation of the splicing seam greatly for improving stability. In addition, the direct use of the splicing seam of the last frame also shortens the calculation time of splicing seam and improves algorithm efficiency.

The invention relates to an artifact-removing image splicing method based on image depth layering. The method comprises the following steps: according to a depth image clustering result, carrying out quantization processing on a depth image of an unspliced image to obtain a depth image mask; classifying the feature points matched with the sift algorithm based on the depth map mask, and calculating respective transformation matrixes of the feature points of different categories; and according to the depth map mask, carrying out depth layering on the spliced image to obtain images of different depth levels, transforming the images of different depth levels by using respective transformation matrixes, and carrying out superposition on the transformed images from deep to shallow according to the depth levels to obtain a final spliced image.

The invention discloses a bionic nerve repair material with a double-layer structure. The bionic nerve repair material is formed by compositing a skin layer and a core layer; the core layer is of a multichannel columnar structure and is prepared from a chitosan solution and collagen gel or chitosan-collagen gel with a gradient condensation method; the skin layer is prepared from high polymer fiberthrough cored weaving and coating; or the skin layer is prepared from the high polymer fiber through electrostatic spinning; or the skin layer is prepared from the chitosan solution and collagen gelor chitosan-collagen gel with an extrusion method. The core layer of the nerve repair material has good cell compatibility, the multichannel structure is more beneficial to growth of nerve cells, thesuture performance and radial compression performance of the material can be improved by the aid of the skin layer, and clinical operation is facilitated.

The invention discloses a method and a device for processing an image based on a vehicle in order to prevent the fact that the splicing can be affected by parallax of an obstacle. The method comprises steps of detecting each splicing public area through a radar adjusting a corresponding preset splicing line of the splicing public area when detecting the obstacle on the preset splicing line of the splicing public area; and adopting an adjusted splicing line to execute image splicing to generate a panorama splicing image. Through the invention, the splicing effect is improved and accurate information surrounding the vehicle is provided through adjusting the splicing line and eliminating the effect of the obstacle on the splicing.

The invention relates to a signal source zooming method and system for a display window. The signal source zooming method for the display window comprises the following steps of obtaining window coordinate parameters of the display window, signal coordinate parameters of a signal source, and intersection coordinate parameters of display units in the display window and the display window; accordingto the window coordinate parameters, the signal coordinate parameters and the intersection coordinate parameters, identifying intersection input data corresponding to the display units at the signalsource in the signal source; and according to a ratio of the signal coordinate parameters to the window coordinate parameters, performing zooming on the intersection input data. Related image parameters corresponding to the display units in the display window obtained after zooming can be seamlessly spliced, so that the problem of discontinuousness at a splicing seam between two adjacent display units is effectively avoided and the image display effect of the corresponding display window can be improved.

The invention belongs to the technical field of bio-based materials, and particularly relates to an artificial biological dura mater and a preparation method thereof. The artificial biological dura mater is prepared from the following raw materials of bacterial cellulose and collagen according to a mass ratio of (0.1 to 3):10; any crosslinking agent is not added, so that the adverse effect to therepair process of the dura mater by the crosslinking agent is reduced, and the clinical use safety of the dura mater product is improved. The artificial biological dura mater has the advantages that the artificial biological dura mater is prepared by an electrostatic spinning method; the pH (potential of hydrogen), temperature and other technology conditions are strictly controlled, so that the good biocompatibility and mechanical property are realized, the requirement of operation suture in the clinical use process is met, and the leakage of cerebrospinal fluid is effectively prevented.

The invention discloses a bridge bottom surface detection device. The bridge bottom surface detection device comprises two support rails which are fixed to a bridge, wherein a working platform is connected between the two support rails in a sliding way; the working platform is connected with a detection unit in which at least detection robots coordinate with one another as a group in a sliding way; the detection robots comprise walking motion parts; the walking motion parts are rotationally connected with detection working parts through extension arms; cameras are arranged on the detection working parts; and the working platform and the detection robots are connected with a remote data center in a wired or wireless manner. Through adoption of the bridge bottom surface detection device andmethod, more than three detection robots are controlled to take photos of box girders and splice the photos simultaneously under the condition of constant shooting posture, so that the splicing difficulty of the taken photos is lowered greatly, and the splicing efficiency and effect are improved. The shooting postures of the cameras can be adjusted according to different conditions through an adjustable extension arm structure.

The invention relates to a preparation method for tubular bacterial cellulose, which adopts double layers of oxygen permeation materials to form a hollow mould by nesting for culturing bacterial cellulose. Based on the disclosed technology, the pressure of internal pressure gas of inner tube of a hollow oxygen permeation mould, external gas of an outer tube of the hollow oxygen permeation mould, and oxygen volume concentration are precisely adjusted, so that the inner tube and the outer tube of the hollow oxygen permeation mould can generate periodic alternating changing of shrinkage and expansion along the radial direction. The radial deformation amplitude is small and does not exceed 5%, and so the cultivation process is a quasi-static process. Based on the prior art, the preparation method provided by the invention doubles the effective region in the bacterial cellulose, shortens the cultivation period of the tubular bacterial cellulose, enables the outer wall of the tubular bacterial cellulose to be more compact, improves the sewing performance of the material and reduces the probability that the tubular bacterial cellulose generates intimal hyperplasia and thrombus at the sewn part. Besides, the preparation method has a short preparation period, a green and environment-friendly preparation process, simplicity and quickness, and low preparation cost.

The invention provides a multi-splicing mode system and method of a panoramic camera. The system comprises a communication module for establishing the communication connection between the panoramic camera and an intelligent terminal; a client control module which is connected to the communication module and is used for selecting and outputting a splicing mode corresponding to the current scene in different application scenes; a camera master control module which is connected to the communication module and is used for receiving the splicing mode message sent by the client end control module and explained as execution command, and emitting control signal; a dynamic intelligent splicing module which is connected to the camera master control module and is used for receiving the control signal from the camera master control module and calculating the splicing distance parameter under the splicing mode so that the real-time in-camera splicing can be conducted by means of the intelligent dynamic stitching technology based on the optimal fusion center line. The invention is advantageous in that in different application scenes, the system can freely and rapidly switch the splicing mode of the corresponding scene, so that the double image of splicing and vagueness under different scenes can be removed.

The invention relates to a synchronous online splicing method based on reference frame conversion and splicing size self-adaptation. To meet the requirement that a splicing result is synchronous and consistent with video flow requirements, a shared memory technique is adopted, it is guaranteed that a frame to be spliced is a currently newest frame, and accordingly synchrony is achieved. To guarantee accuracy of the splicing result, a reference frame conversion technique is used, accumulative errors are reduced, and the splicing duration of video flow is increased. In order to solve the problems that a distributed splicing result size is fixed, memory resources are wasted and splicing performance is reduced, the splicing size self-adaptation technique is adopted, a needed splicing result size is calculated dynamically, unnecessary memory expenses are reduced and splicing performance is improved.

The invention discloses an aerial remote sensing image splicing method assisted by pose information, and relates to the field of aerial photography remote sensing and image splicing. Firstly, aerial remote sensing image data and navigation pose data are obtained and preprocessed; then, the image displacement in the air strip is solved according to the pose information, and an adjacent image overlapping region is obtained; image features are extracted in the image overlapping area for matching, and an image homography matrix is solved for image registration; the image registration quality is analyzed, if registration is judged to be invalid, a homography matrix is directly solved according to the camera pose converted from the navigation data, and image registration is carried out; and finally, fusion is carried out by adopting a suture band feathering fusion method, and panorama geographic positioning is carried out according to the pose information. The pose information of the airborne navigation system is used for assisting in aerial remote sensing image splicing, the remote sensing image splicing speed is increased, splicing of images with rare features can be achieved, the remote sensing panorama with geographic information is finally obtained, and the method has high practical value.