211 results about "Vision Disparity" patented technology

The invention discloses a binocular stereoscopic vision matching method combining depth characteristics. The binocular stereoscopic vision matching method comprises: obtaining a depth characteristic pattern from left and right images through a convolutional neural network; calculating a truncation similarity measurement degree of pixel depth characteristics by taking the depth characteristics as the standard, and constructing a truncation matching cost function combining color, gradients and depth characteristics to obtain a matched cost volume; processing the matched cost volume by adopting a fixed window, a variable window and a self-adaptive weight polymerization or guide filtering method to obtain a cost volume polymerized by a matching cost; selecting an optimal parallax error of the cost volume by adopting WTA (Wireless Telephony Application) to obtain an initial parallax error pattern; then finding a shielding region by adopting a double-peak test, left-right consistency detection, sequence consistency detection or shielding constraint algorithm, and giving a shielding point to a parallax error value of a same-row point closest to the shielding point to obtain a parallax error pattern; and filtering the parallax error pattern by adopting a mean value or bilateral filter to obtain a final parallax error pattern. By adopting the binocular stereoscopic vision matching method combining the depth characteristics, the incorrect matching rate of three-dimensional matching can be effectively reduced, the images are smooth and image edges including edges of small objects are effectively kept.

The invention discloses a fast 3D face identifying method based on double-eye passive solid sight, which includes the following steps: 1) a non-contact short shaft parallel binocular stereo vision system is built by applying two high-definition digital cameras; 2) after system calibration is finished, face detection and collection based on a haar-AdaBoost sorting machine is carried out on a preview frame image for obtaining corresponding upper and lower stereoscopic vision graph pairs and estimating a sight difference; image correction is carried out on a face area for obtaining the upper and lower stereoscopic vision graph pairs vertical to the polar lines inside and outside the area; 3) the accurate location on the eyes and a spex nasi is captured by applying a Bayesian and the haar-AdaBoost sorting machines as well as point cloud 3D information for building a benchmark triangle; 4) the corresponding sub pixels in the middle and small areas are matched by applying the pyramidal parallel search solid graph of a phase relevant arithmetic based on a complex wavelet; 5) pose normalizing and hole filling are carried out on the faces under different poses by applying the built benchmark triangle; 6) expression normalization is carried out on different faces based on the suppose that the surface geodesic distance of the face is invariable; 7) the 3D faces after normalization are identified by utilizing the arithmetic. The method has the beneficial effects of: mainly solving the problems of being hard to fast and automatically obtain the passive stereoscopic vision and identifying the 3D point cloud information of the dense and accurate face under different poses and expressions, thus leading the 3D face identifying process to be faster, more hidden, safer and more reliable.

A system, device within a system, and a method for enhancing visual perception in a two-dimensional image is disclosed, providing the induction of retinal disparity in the viewer by a device with which the two-dimensional image is viewed. With fusion of the retinally disparate images of the two-dimensional image, the viewer experiences the enhancement of his or her visual perception in the two dimensional image, which includes enhancement of depth perception and enhancement in clarity perception.

The invention relates to a binocular vision positioning method for target grabbing of an underwater robot, and belongs to the field of computer vision. The method is mainly used for accurately acquiring three-dimensional information of a grabbed target when an underwater robot works. The method comprises the following steps: double-target positioning: calculating internal and external parameters of left and right cameras; target detection: positioning a target object detection frame; binocular image correction: carrying out distortion correction and stereo correction, and determining a right image target area; binocular image stereo matching: extracting image feature points, describing the feature points, performing stereo matching, and removing mismatching; and calculating the three-dimensional information of the target in the image under the left camera coordinate. According to the method, the accurate parallax value is obtained by extracting feature points, removing unstable featurepoints through non-maximum suppression, constructing a binary descriptor, matching the feature points and removing mismatching. Through the scheme, the binocular stereo matching robustness can be improved, and meanwhile, the three-dimensional information of the detection target can be accurately obtained, so that the real-time positioning requirement on the target when the underwater robot grabsthe target is met.

The invention discloses a binocular visual sense mileage calculating method based on image gradient combined optimization; the method comprises the following steps: preprocessing an inputted binocularvideo, and calculating a binocular video frame disparity map so as to obtain a tolerance distance; zooming to form multi-scale images and forming a pyramid model, and calculating features to obtain aseries of feature images; using a Kalman filter to process the image according to the tolerance distance, and predicting and estimating a camera attitude motion process, wherein a camera motion modelis built in the Kalman filter; using a binocular visual sense navigation algorithm based on gradient to calculate the current frame accurate camera attitude; using the current frame camera attitude to update the camera motion model in the Kalman filter. The invention proposes the optimization algorithm combining two gradient expansions, and creatively uses the image gradient as features, thus effectively preventing outdoor illumination change influences; the method uses a plurality of key frames so as to optimize the camera attitudes, is good in performance, and can simultaneously carry out dense three dimensional reconstructions.

The invention discloses a method and a system based on binocular stereoscopic vision for passenger flow density estimation. The system comprises binocular parallel cameras, a DSP (Digital Signal Processor), a communication module, a video output module and the like, and the system is adopted for estimating passenger flow density. The method comprises the following steps: (1) calibrating a left camera and a right camera to obtain the internal and external parameters of the left camera and the right camera; calculating the accurate principal point difference of the left camera and the right camera; (2) collecting a left image and a right image in real time and carrying out position correction, and adopting a self-adapting window stereo matching method to obtain a current frame parallax error image; (3) splitting a foreground human body target to obtain a binary foreground human body target image; (4) generating a dimensional mapping image of the foreground human body target; (5) calculating the degree of congestion, and obtaining the density estimation through the non-linear relationship between the degree of congestion and the number of people. The method is independent of the influences of scene illumination change, shadow, perspective effect and blocking, and the system has the characteristics that the equipment is simple and the accuracy rate of the density estimation is high.

The invention discloses a borderless terminal display device. The borderless terminal display device comprises a display screen and a glass panel sticking to the upper part of the display screen, wherein the display screen comprises a main screen body in the center and bent screen bodies extending from two transverse sides of the main screen body in a bending way; the pixel points of the protections, on the plane where the main screen body is positioned, of the bent screen bodies are as large as those of the main screen body; the glass panel comprises a main cover plate in the center and bent cover plates extending from two transverse sides of the main cover plate in a bending way; the main cover plate corresponds to the main screen body; the bent cover plates correspond to the bent screen bodies; the inner surfaces of the bent cover plates correspond to the outer surfaces of the bent screen bodies. The borderless terminal display device provided by the invention can achieve real borderless display, and has no display vision disparity when being used by a user.

The invention discloses a terminal screen display parameter adjustment method comprising the steps that eyeball identification information is acquired, and a focusing area corresponding to the eyeball identification information is confirmed in a screen display interface; and screen display parameters of all areas of the screen display interface are adjusted respectively according to coverage scope of the focusing area so that vision disparity of the screen display interface can be realized. The invention also discloses a terminal screen display parameter adjustment device. Use duration of a terminal battery can be effectively guaranteed, and vision disparity of the terminal display interface can be realized so that the terminal display effect can be enhanced.

The invention relates to an edge-based parallax image calculation method applied to binocular stereo vision. The method includes: acquiring gradient images of a scene on the basis of depth images, shot by two depth cameras, of the same scene; acquiring a minimum matching cost value and a second minimum matching cost value of first pixel points in the gradient images according to a difference cost function; and determining that parallax of the first pixel point is invalid parallax if a ratio of the minimum matching cost value to the second minimum matching cost value is greater than a preset ratio. According to the embodiment of the invention, a low texture region or a repeated texture region corresponding to the first pixel point can be determined in the case where the ratio of the minimum matching cost value to the second minimum matching cost value is greater than the preset ratio, then it is determined that the parallax of the first pixel point is the invalid parallax, thus calculating parallax for pixel points outside high-level texture regions can be reduced, accurately calculate the parallax is more facilitated, and then an edge-based stereo matching parallax image is output on the basis of the parallax.

The invention discloses a depth estimation method based on binocular vision and laser radar fusion. The method comprises the following steps: registering data acquired by a laser radar and a binocularcamera through joint calibration; obtaining a laser radar disparity map according to a joint calibration result; obtaining a binocular disparity map through a binocular stereo matching algorithm, performing confidence analysis on the binocular disparity map, removing points with low confidence in the binocular disparity map, and obtaining the binocular disparity map after confidence processing; performing feature extraction and fusion on the obtained laser radar disparity map and the binocular disparity map after confidence processing; carrying out further feature extraction and parallax regression through a cascade hourglass structure; adopting relay supervision to utilize output of front and back cascaded hourglass structures; and outputting an accurate and dense disparity map after fusion. According to the method, a more effective network structure is designed, the features of the laser radar disparity map and the binocular disparity map are better extracted and fused, and a more accurate disparity map is obtained.

Described are methods and apparatus for adjusting images of a stereoscopic image pair based on keypoint matches. The quality of the key point matches is first evaluated to determine whether the quality exceeds a keypoint quality threshold. If the quality level of the keypoint matches exceeds the threshold, the vertical disparity between the images of the stereoscopic image pair can be evaluated based on vertical disparity vectors between the keypoint matches. If the vertical disparity is below a threshold, no adjustment of the stereoscopic image pair may be performed. If the vertical disparity is above the threshold, an affine correction may compensate for pitch, roll, and scale differences between the images. A projective correction may compensate for yaw differences. The vertical disparity between the two images is then evaluated after the corrections to determine if additional adjustment should be performed.

The invention discloses a serial channel fusion pedestrian detection method based on binocular vision. According to the method, a parallax image is generated according to a binocular RGB image, a convolutional neural network is constructed, an RGB image channel and a parallax image channel are connected in series to serve as input of the convolutional neural network, the complementary relation between the RGB image and the parallax image is learnt, a part with a predicted value greater than a set value is selected from a detection result output by the convolutional neural network, non-maximumsuppression is performed on the part, and a final pedestrian position is obtained. Through the method, since the RGB color image has rich color information and the parallax image is invariant to illumination, pedestrian detection can be more robust and accurate by combining the RGB image and the parallax image.

The invention provides a flexible touch screen and a manufacturing method thereof. A touch induction electrode layer (2) and wires thereof and a touch drive electrode layer (4) and wires thereof are manufactured on the upper surface and the lower surface of a same flexible substrate (1) respectively, therefore, the technology complexity of the flexible touch screen can be reduced, the thickness of the flexible touch screen can be decreased, and the flexibility of the flexible touch screen is improved; the flexible substrate (1) is provided with via holes (7), the wires on the surface of one side are led to the other side, and then a flexible printed circuit (6) is welded to the surface of the same side; by improving patterns of the touch induction electrode layer (2) and additionally arranging shielding electrodes (22), vision disparity caused by the patterns of the touch induction electrode layer (2) is reduced, and the visual effect and the technology stability in the etching process are improved.

A quality detection method for a liquid crystal screen comprises: acquiring an optical parameter of a liquid crystal screen; collecting an image of the liquid crystal screen; and if it is determined that the optical parameter is not in a preset optical parameter range and / or the collected image of the liquid crystal screen is inconsistent with a pre-stored image, determining that the liquid crystal screen is unqualified. By means of the method, unqualified types of products can be effectively distinguished, and unqualified locations of products can be recorded, thereby effectively reducing misjudgment or missed inspection due to vision disparity of human, so as to improve the outgoing quality and yield of the products. Further provided is a quality detection apparatus for a liquid crystal screen.

The non-reference three-dimensional image quality evaluation method based on the fused image and the enhanced image comprises the following steps: firstly, performing red, green and blue three-channel fusion on left and right viewpoints of the three-dimensional image based on binocular fusion, binocular competition, binocular inhibition and other characteristics of a human vision system to obtain a color fused image; secondly, obtaining a disparity map of the distorted stereo image pair by using a stereo matching algorithm, and weighting a gray-scale map of the color fusion image by using a gradient weight of the disparity map; thirdly, generating an enhanced image according to the fused image and the disparity map; thirdly, extracting natural statistical characteristics from the fused image and the enhanced image in a spatial domain, and extracting kurtosis and skewness characteristics from the disparity map; and finally, fusing the extracted features, and sending the fused features into support vector regression (SVR) to obtain the quality of the to-be-evaluated stereo image.

The invention discloses a liquid crystal screen defect and dust distinguishing method based on a binocular visual system and a detection device. The distinguishing method comprises the following stepsthat the binocular visual system is calibrated by using a target frame with a liquid crystal layer acting as a reference surface, and the mapping relation between the image coordinate systems corresponding to the binocular visual system is determined; imaging of the liquid crystal screen is performed by using the binocular visual system so that a binocular image A and an image B are obtained; image processing is performed on the image A and the image B so as to acquire defect information in the images; the image A and the image B are converted to a reference surface coordinate system; as forthe defects in the image A and the image B, the parallax information of the defects on the liquid crystal screen in the image A and the image B is acquired in the reference surface coordinate system;and the parallax information of the defects in the image A and the image B is judged, the defects are the foreign matter defects if the parallax information is the same and the defects are the surfacedust if the parallax information is different. The problems in the prior art that the foreign matter defects and the surface dust cannot be distinguished by automatic detection equipment can be solved.

The invention provides an object detection method and a device. The object detection method comprises steps that: disparity images are acquired from a stereo camera for object area imaging; planes are detected from the disparity images; object surface candidates are selected from the planes detected from the disparity images according to characteristics of a physical surface of an object; the object surface candidates are converted into front view surfaces, and the front view surfaces are object surface images acquired by utilizing the stereo camera to shoot from a visual angle just pointing to the physical surface of the object; the object is detected at least partially on the basis of the front view surfaces after conversion. According to the method and the device, influence of original deformation can be eliminated, the object is detected by utilizing already-extracted non-deformation characteristics and / or a classifier and on the basis of the front view surfaces after conversion.

The invention provides a binocular image super-resolution method based on multi-scale feature fusion. Firstly, a feature extraction module adopts mixed jump type residual connection, pooling blocks of an improved void space pyramid form a loopback structure for extracting multi-scale features of an image, and then an expansion residual and the loopback structure are alternately cascaded to fuse the extracted features; then, a parallax attention module is introduced to obtain a corresponding relation in the binocular image, useful information of an integrated image pair is integrated, and a transitional expansion residual block is used to learn the network capability of stereo matching features to obtain a parallax image of the binocular image; and finally, mapping a low-dimensional space to a high-dimensional space by using four expanded residual blocks, reconstructing a super-resolution left (right) graph through sub-pixel convolution, and applying FReLU to the whole network to improve the efficiency of capturing spatial correlation. According to the method, the multi-scale features of the image are extracted by using the expansion residual loop structure, the excellent super-resolution performance is realized, and the method has wide applicability.

The invention discloses a method for detecting the overall dimension of a running vehicle based on binocular vision. The method comprises the following steps: calibrating and correcting a binocular camera; performing moving object recognition and tracking on the corrected view to obtain a vehicle feature region; enabling the identified vehicle surface to be subjected to texture enhancement processing, so that the problem of low detection precision of a weak texture surface is solved; based on vehicle driving scene characteristics, providing a stereo matching algorithm based on time sequence propagation to generate a standard disparity map, and improving the vehicle overall dimension measurement precision; performing three-dimensional reconstruction on the generated disparity map to generate a point cloud map; providing a space coordinate fitting algorithm, fitting a plurality of frames of point cloud images of the tracked vehicle, generating a standard vehicle overall dimension image. The problem that the overall dimension of the vehicle cannot be completely displayed through a single frame of point cloud image is solved. The method is not limited by the vehicle speed in measurement effect, high in measurement precision, wide in measurement range and low in cost. The binocular camera has the advantages of being flexible in structure, convenient to install and suitable for measurement of all road sections.

The invention provides an underwater environment three-dimensional reconstruction method based on binocular vision, and the method comprises the following steps: 1, collecting and obtaining an underwater image, carrying out the underwater calibration of a binocular camera, and obtaining the needed related parameters of the binocular camera; 2, preprocessing the collected underwater image, including image denoising, image enhancement, image sharpening, image restoration and underwater image defogging; 3, performing feature detection on the preprocessed binocular image in the step 2, and performing stereo matching by using an improved Census and NCC fused stereo matching algorithm to obtain a disparity map containing depth information; and 4, performing three-dimensional reconstruction on the disparity map in the step 3 by using a PCL three-dimensional reconstruction method introducing a moving least square method, and restoring an underwater three-dimensional environment in the image. The invention introduces the moving least square method to solve the problems of point cloud discretization and point cloud vulnerability, thus visually reflecting a three-dimensional effect by a processing effect from multiple angles, and restoring the underwater three-dimensional environment.

The invention relates to a stereoscopic-vision-based rapid workpiece sorting method and device. A conveyor belt is arranged on a working platform. A binocular camera of which the base line is parallelto the moving direction of the conveyor belt is arranged above the working platform. The side portion of the working platform is provided with a sorting module comprising a mechanical arm. A controller is connected with a displayer, the sorting module, the conveyor belt and the binocular camera. An ROI area is determined through a binocular image. Coordinates and a parallax error of a workpiece in a depth map are obtained through stereo matching, and time consumption is reduced. According to the conveying speed of the conveyor belt and the time interval of former and latter frames of images,the corresponding position of the workpiece in the latter frame of the image is calculated on the basis of the position of the workpiece in the former frame of the depth map. The error is corrected and the correctness is improved through stereo matching at intervals of M frames. Through the method and the device, the stereo matching area is reduced; the calculated quantity is small; there is no need to conduct stereo matching on each frame of the image; the time consumption is reduced; the practicality is high; stereo matching of sorting can be rapidly implemented on relatively low-price hardware equipment with poor performance; the hardware dependency is reduced; and compared with a similar hardware platform, the sorting success rate is increased, and the production efficiency is high.

The invention discloses a suspension preview control method based on a binocular vision technology. The suspension preview control method comprises the following steps: S1, acquiring a pavement image in a driving direction in front of a vehicle in real time; S2, identifying the road surface image by using a target detection algorithm, and judging whether a deceleration strip or pit instantaneous road surface excitation exists or not; S3, obtaining a parallax error by combining a stereo matching algorithm with the position information of the deceleration strip or the pit; S4, estimating the distance of the deceleration strip or the pit according to a binocular distance measurement algorithm in combination with the internal and external parameters of the binocular camera and the parallax value, and calculating the time when the vehicle arrives at the deceleration strip or the pit by the vehicle ECU according to the vehicle speed information; S5, designing a suspension preview controller according to a model predictive control principle, and switching to the suspension preview controller in advance before the estimated time when the vehicle arrives at the deceleration strip or the pit. The invention further provides a suspension control device. According to the invention, the arrangement condition of the suspension of the vehicle is matched with the driving condition of the vehicle, and the driving comfort and safety of the vehicle are improved.

The invention belongs to the technical field of image processing, and discloses an unsupervised monocular view depth estimation method based on multi-scale unification, and the method comprises the following steps: S1, carrying out the pyramid multi-scale processing of an input stereo image pair; s2, constructing a network framework for encoding and decoding; s3, transmitting the features extracted in the encoding stage to a reverse convolutional neural network to realize feature extraction of input images of different scales; s4, performing unified up-sampling on the disparity maps of different scales to an original input size; s5, performing image reconstruction by using the input original image and the corresponding disparity map; s6, constraining the accuracy of image reconstruction; s7, training a network model by adopting a gradient descent method; and S8, fitting a corresponding disparity map according to the input image and a pre-training model. According to the method, networktraining does not need to be supervised by using real depth data, the binocular image which is easy to obtain is used as a training sample, the obtaining difficulty of network training is greatly reduced, and the problem of depth map holes caused by low-scale disparity map blurring is solved.

The invention discloses a method for automatically detecting obstacles based on binocular vision. The method comprises the following steps: receiving image information from two paths of binocular vision cameras, the image information including a V-disparity map and a U-disparity map; calculating a conversion formula between image coordinates of the image information and coordinate systems according to the coordinate systems of the two paths of cameras; comparing image information that identical time codes of the received image information correspond to, and obtaining a disparity map; and obtaining related lines of the disparity map according to the disparity map, and obtaining the position information of obstacles according to the conversion formula and the related lines of the disparity map. The method provided by the invention can utilize binocular vision to simulate eyes of humans and the process of stereoscopic perception of humans, and automatically detect obstacles for road driving, and the method has strong adaptability, good real-time performance, and good concealing performance.

The invention provides a target positioning method and device, electronic equipment and a storage medium, and the method comprises the steps: calculating a disparity map based on a left view and a right view, which are captured by a binocular camera and contain a to-be-positioned target; inputting the left view into the trained deep learning network, and outputting a target mask in the left view;and based on the disparity map and the target mask in the left view, calculating three-dimensional space coordinates of the to-be-positioned target by using a three-dimensional reconstruction projection method. According to the invention, binocular stereo vision and deep learning are combined; the binocular camera is utilized to calculate position deviation between corresponding points of the leftview and the right view according to a triangulation principle; specific target recognition processing is carried out on the image by using a deep learning method, real-time positioning is carried out on a scene target in combination with three-dimensional reconstruction information on the basis of target recognition, a target positioning process is simplified, primary and secondary targets do not be distinguished, and all target positions in a view field are calculated at the same time; deep learning may expand the range of target positioning applications for specific targets and ordinary targets.

The invention belongs to the field of robot perception, and particularly relates to a disparity map enhancement method based on RGB and DVS image fusion in a high dynamic range scene. The method comprises the following steps: S1, deploying a binocular RGB camera and a DVS camera, and calibrating the binocular RGB camera and the DVS camera; s2, collecting an RGB image and a DVS image of a binocularcamera in a scene, and performing multi-scale weighted fusion after registration; s3, generating an HDR image for computer vision for the fused image; and S4, based on the HDR image generated in thestep S3, using an improved binocular stereo matching algorithm SGM to generate a disparity map. In a scene with a large imaging dynamic range such as a tunnel, the problems of under exposure and highexposure of a camera are solved, the quality of a generated image is improved, and meanwhile, for the problems of discontinuity and instability of an image edge region, edge detail information is enriched as much as possible by introducing other information sources, and the accuracy of the finally generated disparity map at the edge of the image is improved.

The present invention provides a three-dimensional information acquisition method and device based on line laser and binocular vision. The method includes the following steps that: line laser emittedby a line laser is controlled to be projected onto a measurement object; the left camera and right camera of a binocular camera are utilized to measure the left image and right image of the measurement object respectively, the parameters of the left camera and the parameters of the right camera are identical, in the left image and right image, distances between the measurement object and the edgesof the images are smaller than a first set value; stereoscopic correction is performed on the left image and the right image, so that the corrected left image and the corrected right image are be inline alignment with each other, or the reprojection distortion of the left image and the right image is smaller than a second set value; the corrected left view and the corrected right view are matched with each other, so that line laser imaging matching point pairs can be obtained; and left and right view parallax is obtained according to the line laser matching point pairs, and the three-dimensional information of the measurement object is calculated according to the left and right view parallax.

The invention provides a touch substrate and a touch display device, relates to the technical field of display and is used for solving the problem that the display effect is reduced due to vision disparity which is caused by the fact that the area covered with electrode patterns reflects light and the area without electrode patterns doesn't reflect light in the touch substrate. The touch substrate comprises a plurality of electrodes, and at least one electrode is in a hollowed structure. The touch substrate is applied to the display device, so that the display device has a touch function.

The invention belongs to the technical field of vehicle accessories, and particularly relates to an outside rear-view mirror and a method for adjusting the outside rear-view mirror when used for a vehicle. The outside rear-view mirror comprises a reflective mirror set and a shell. The reflective mirror set is composed of a first reflective mirror and a second reflective mirror. The first reflective mirror and the second reflective mirror are vertically arranged in the shell in the vertical direction. A drive mechanism is located in a cavity formed between the reflective mirror set and the shell and connected with the first reflective mirror and the second reflective mirror to drive the first reflective mirror and the second reflective mirror to rotate in the shell. Both the first reflective mirror and the second reflective mirror are convex lenses. The curvature radius of the first reflective mirror is the same as that of the second reflective mirror. By the adoption of the outside rear-view mirror, the dead zone phenomenon of a conversional outside rear-view mirror can be greatly and effectively avoided, vision disparity is reduced, and erroneous judgment is reduced; and the method for adjusting the outside rear-view mirror is reliable, practical and suitable for being widely applied and popularized as a vehicle accessory.

In a stereo vision system comprising two cameras shooting the same scene from different positions, a method is performed for determining dense disparity fields between digital images shot by the two cameras, including the steps of capturing a first and a second image of the scene, and determining, for each pixel of the second image, the displacement from a point in the first image to such pixel of the second image minimising an optical flow objective function, wherein the optical flow objective function includes, for each pixel of the second image, a term depending in a monotonously increasing way on the distance between the epipolar line associated with such pixel and the above point in the first image, such term depending on calibration parameters of the two cameras and being weighed depending on the uncertainty of the calibration data.