943 results about "Perspective transformation" patented technology

In some arrangements, product packaging is digitally watermarked over most of its extent to facilitate high-throughput item identification at retail checkouts. Imagery captured by conventional or plenoptic cameras can be processed (e.g., by GPUs) to derive several different perspective-transformed views—further minimizing the need to manually reposition items for identification. Crinkles and other deformations in product packaging can be optically sensed, allowing such surfaces to be virtually flattened to aid identification. Piles of items can be 3D-modelled and virtually segmented into geometric primitives to aid identification, and to discover locations of obscured items. Other data (e.g., including data from sensors in aisles, shelves and carts, and gaze tracking for clues about visual saliency) can be used in assessing identification hypotheses about an item. A great variety of other features and arrangements are also detailed.

An electronic device for providing a display that changes based on the user's perspective is provided. The electronic device may include a sensing mechanism operative to detect the user's position relative a display of the electronic device. For example, the electronic device may include a camera operative to detect the position of the user's head. Using the detected position, the electronic device may be operative to transform displayed objects such that the displayed perspective reflects the detected position of the user. The electronic device may use any suitable approach for modifying a displayed object, including for example a parallax transform or a perspective transform. In some embodiments, the electronic device may overlay the environment detected by the sensing mechanism (e.g., by a camera) to provide a more realistic experience for the user (e.g., display a reflection of the image detected by the camera on reflective surfaces of a displayed object).

The focal length estimation method and apparatus claimed in this application aligns plural overlapping images with one another for constructing an image mosaic. This is accomplished by computing a planar perspective transformation between each overlapping pair of the images, computing from the planar perspective transformation a focal length of each image of the pair, computing from the focal length of each image a focal length transformation, computing a rotational transformation for each of the pair of images whereby a combination of the rotational transformation and the focal length transformation relates the respective image to a three-dimensional coordinate system. Registration errors between the pair of images are reduced by incrementally deforming the rotational transformation of one of the pair of images. The planar perspective transform is a matrix of warp elements, and the focal length is computed as a function of the warp elements, the function being derivable by constraining a first two rows or the first two columns of the matrix to have the same norm and to be orthogonal. The focal length of one image of a pair of images is found by applying the constraint on the matrix columns, while the focal length of the other image of the pair is found by applying the constraint on the matrix rows

In some arrangements, product packaging is digitally watermarked over most of its extent to facilitate high-throughput item identification at retail checkouts. Imagery captured by conventional or plenoptic cameras can be processed (e.g., by GPUs) to derive several different perspective-transformed views—further minimizing the need to manually reposition items for identification. Crinkles and other deformations in product packaging can be optically sensed, allowing such surfaces to be virtually flattened to aid identification. Piles of items can be 3D-modelled and virtually segmented into geometric primitives to aid identification, and to discover locations of obscured items. Other data (e.g., including data from sensors in aisles, shelves and carts, and gaze tracking for clues about visual saliency) can be used in assessing identification hypotheses about an item. Logos may be identified and used—or ignored—in product identification. A great variety of other features and arrangements are also detailed.

In a virtual three-dimensional space, at least multiple player characters and a movable special object exist. The multiple player characters move in the virtual three-dimensional space according to operations of each player. The special object can be moved in the virtual three-dimensional space according to operations of each player character. An image, which is perspective-transformed in a state that a visual axis of a virtual camera is directed to the direction of the special object, is displayed as a game screen. An area, which is within a predetermined distance from the special object, is fixed as a safety zone. Hit points of the player characters, which are within the safety zone, do not decrease. However, hit points of the player characters, which are outside the safety zone, decrease.

A wide-angle image automatic joining method comprises: a step one of setting a moving speed of a wide-angle camera and an image collecting period; a step two of obtaining a superposition pixel width of two adjacent images; a step three of building an image non-linear aberration correcting mathematical model; a step four of creating an image data coordinate conversion relationship; a step five of calculating an aberration vector parameter and an outer parameter value; a step six of performing an inversion operation on the image with aberration to obtain the whole real image after correcting the non-linear aberration; a step seven of rapidly finding common characteristics of the two adjacent images in the superposition width of the images with corrected non-linear aberration; a step eight of realizing registration of the two images by perspective conversion; a step nine of joining the two registered images; a step ten of inosculating the joining position of the images based on a wavelet bicubic interpolation searching method. According to the invention, continuous inosculation and seamless join of the adjacent images can be automatically realized.

A 360 degree panoramic aerial view generation method used for assisting vehicle driving is disclosed. The method is characterized by: firstly, erecting four wide-angled cameras around a vehicle and guaranteeing that the cameras can cover all the field areas surrounding the vehicle; secondly, calibrating a camera parameter and carrying out distortion rectification on the collected image; then, carrying out areal perspective transformation on the rectified image so as to obtain the aerial view with a downward view effect; finally, establishing a image splicing lookup table so that a system can rapidly complete splicing of a panoramic picture through a lookup table method. The method is simple is easy to perform. A hardware equipment requirement is low. The system realizes the splicing of four pictures through the lookup table mode, which is rapid and has less amount of calculation. The operation parameter which is calibrated once by using the method can be served as a constant, be stored and be used in real time hardware system work so that system processing time can be substantially reduced and the real-time performance requirement of the system can be satisfied.

The invention discloses a panorama parking calibration method and a device, and an automatic calibration method, wherein the panorama parking calibration method comprises the steps of obtaining pictures which is captured by an automobile camera; using default parameters of a panorama parking system to generate an aerial view; distinguishing tag blocks on the aerial view; distinguishing controlling points according to the tag blocks; obtaining actual positions and ideal positions of the controlling points, wherein the controlling points are intersection points of lines and distributed in a net shape, and the ideal positions are appointed positions in the aerial view; and optimizing by utilizing difference between the actual positions and the ideal positions of each controlling point in the aerial view and obtaining lens distortion correction parameters and perspective transformation parameters of a wind-angle camera which corresponds to the aerial view. The area of the tag blocks is greater than that of the intersection points and is easy to distinguish. A light color is adopted on distinguish strips and a deep color is adopted on a carrier so as to reduce effects which are caused by reflect light, an absolute coordinate of a calibration pattern on a coordinate system is positioned and calibrated accurately, and a success rate of calibration is improved.

The invention provides a vehicle line walking method and system based on machine vision. The line walking method comprises the following steps: an image acquisition module acquires an RGB color image of a road surface; an image processing module converts the RGB color image into a grayscale image; a binarization module obtains a binarization image to separate a lane line; an edge detection module obtains an edge image containing an inner edge and an outer edge of the lane line; a processor acquires parameters of the lane line to establish a lane line model, acquires lane position data of a vehicle in a world coordinate system through inversely perspective conversion, and determines a driving mode of the vehicle according to road surface image information; parameter adjustment of a steering engine and a motor can be realized by using acquired vehicle corner parameters and distance parameters, adopting the driving mode of the vehicle, and utilizing a sectional adaptive control strategy so as to realize real-time control of driving directions and driving speed of the vehicle. According to the method and system, provided by the invention, the line walking speed in the driving process of the vehicle can be increased, the line walking can be completed within the time of 10 milliseconds, and the stability in the driving process of the vehicle is strong.

The invention relates to a traffic signal light intelligent control system and a control method thereof, which belong to the technical field of intelligent traffic; the system comprises four CCD camera heads, a signal processor and a controller; and the method comprises the following steps: establishing a coordinate system and the transformation relation thereof; sequentially carrying out distortion correction, sharpening and perspective transformation on a traffic flow image; extracting edges of a vehicle image and filtering out lane lines and traffic sign arrows in the image; cutting the image; carrying out edge projection for obtaining pixel coordinates which correspond to the queue length of vehicles waiting to pass on each type of lane in each direction under the computer image coordinate system; calculating the actual queue length of the vehicles waiting to pass; obtaining time required for leading all the vehicles on each type of the lane in each direction to pass; and controlling the opening and the closing of a traffic signal light according to the time of leading all the vehicles to pass. The traffic signal light intelligent control system can control the opening time and the closing time of the traffic signal light in a high-efficient and energy-saving manner, thereby providing a scientific information base for realizing the urban intelligent traffic.

A chromokeying system includes a store that contains a representation of a known coded chromokeying pattern. The system further includes a frame grabber that grabs a video image including at least a portion of the known coded chromokeying pattern and a foreground subject. A perspective transformation computation unit computes a perspective transformation by comparing the video image of the portion of the coded chromokeying pattern with the known chromokeying pattern. The system may further include a background generation unit that receives a background image and applies the perspective transformation to the background image to produce a transformed background. The portion of the coded chromokeying pattern in the video image can then be replaced by the transformed background to create a composite image.

The invention provides a road marking line extracting method based on a forward camera head in automatic driving. The method comprises steps of obtaining a normal incidence view of a road marking line in front of a current vehicle via reverse perspective transformation; restraining an initial threshold value of current solving road segmentation according to parts of prior knowledge formed based on normal running modes of vehicles; by using the initial threshold value, segmenting the normal incidence view; judging the parallel property presented by edge pixel points and the equidistant property presented by a projection histogram; and extracting road marking lines including lane lines, road traffic signs, vehicle stop markings and zebra marking lines. According to the invention, execution speed is fast; by operating the projection histogram, effects on detection and extraction imposed by incomplete image details can be reduced; robustness of the extracting method is improved; constraints on few prior knowledge are increased; and by adopting the approximate, divide-and-conquer and dimension reduction technological means, the range of the extraction precision is further reduced and processing of sub-regions of parts of road scenes is inhibited.

One of stereo cameras is set such that a front view of a support surface of a workpiece is imaged, an image produced by the camera is displayed, and a range of an area where measurement processing is enabled is assigned by a rectangular frame. A manipulation assigning an upper limit and a lower limit of a height measurement range is accepted. When each assignment is fixed, zero is set as a z-coordinate to each constituent pixel of an image to which the rectangular frame is set, and a z-coordinate based on the upper limit of the height measurement range and a z-coordinate based on the lower limit are set to coordinates corresponding to the rectangular frame. Perspective transformation of three-dimensional information produced by the setting is performed from a direction of a line of sight set by a user, a produced projection image is displayed on a monitor.

The invention discloses a lane line detecting and tracking and detecting method, and relates to the technical field of image processing. The lane line detecting method of the invention comprises: performing an inverse perspective transform processing on a preprocessed image based a homography matrix H, performing a blocking and straight line detection on the processed image, and then making endpoints of straight lines of the upmost image block and the minimum image block as match points to obtain a smooth curve x, and based on weight of the curve x, selecting a curve with the maximum weight as a lane line pair of a current frame image. The lane line tracking and detecting method of the invention comprises: firstly outputting a lane line pair of a first frame image based on the above detecting method, wherein a lane line of a subsequent frame is determined by a target region of a current frame detected by a loop-structured tracker, and a parameter of the tracker is derived from a lane line coordinate of the former frame; outputting a frame by the above detecting method in case of a detection error. The method is applicable for lane line detecting, and the method has strong robustness, great instantaneity and high recognition rate.

The invention discloses a method for the automatic extraction of the position of a license plate in a vehicle monitoring image and perspective correction, and the method comprises the steps: finding an approximate range of the license plate in the image through employing the edge and color information; carrying out the linear detection of the image; determining four boundary lines forming a perspective deformation license plate through employing the regional boundary information and combining with significant straight lines; solving four intersection points of the significant straight lines, and building mapping relation between the four intersection points with four corners of a target rectangle; carrying out the reverse calculation of a perspective correction matrix, and completing the perspective correction of a license plate. The method obtains the positions of four corners of a license plate with perspective distortion, builds the mapping relation between the four corners of the license plate with perspective distortion and the four corners of the target rectangle, calculates a perspective transformation matrix, and completes the perspective correction of the license plate. Compared with an affine transformation method based on horizontal and vertical rotation, the method truly describes the perspective transformation, and guarantees the higher license plate image correction precision. Compared with an affine transformation algorithm, the method can detect the perspective distortion information of the license plate, also can precisely locate the boundary corners of the license plate, calculates the perspective transformation matrix, and accurately achieves the perspective correction of the license plate.

Signal processing devices and methods estimate a geometric transform of an image signal. From a seed set of transform candidates, a direct least squares method applies a seed transform candidate to a reference signal and then measures correlation between the transformed reference signal and an image signal in which the reference signal is encoded. Geometric transform candidates encompass differential scale and shear, which are useful in approximating a perspective transform. For each candidate, update coordinates of reference signal features are identified in the image signal and provided as input to a least squares method to compute an update to the transform candidate. The method iterates so long as the update of the transform provides a better correlation. At the end of the process, the method identifies a geometric transform or set of top transforms based on a further analysis of correlation, as well as other results. Phase characteristics are exploited in the process of updating coordinates and measuring correlation. The geometric transform is used as an approximation of the geometric distortion of an image after digital data is encoded in it, and is used to compensate for this distortion to facilitate extracting embedded digital messages from the image. Due to the errors in the approximation, a signal confidence metric is determined and used to weight message symbol estimates extracted from the image.

The invention discloses an identification method and system of lateral deviation of a vehicle driving route. According to the method and system, firstly, an image, which is shot by a camera mounted ona vehicle, of a road along a vehicle running direction is acquired; an internal-parameter matrix of the camera is utilized to carry out de-distortion processing on the video frame, and inverse perspective transformation is carried out on a selected to-be-transformed region src; then threshold segmentation is carried out on the video frame to separate lane lines from a background region; then locations where maximum values appear are derived by counting to use the same as intersection points of the lane lines and the bottom of the image; and whether correction of lateral deviation of the driving route needs to be carried out is judged according to separation distances between the locations of the intersection points and a preset lateral mounting location of the camera on the vehicle. In acase where the vehicle deviates from a road center too far, the method and system can know the case in time, and remind a driver or carry out correction; and the method and system can accurately detect all of sudden illumination changes, tree shade shading, road surface stains and many other conditions, and are high in applicability, low in costs, high in precision and better in real-time performance and stability.

Provided is a lane line detection method, comprising the following steps: collecting a lane line image including a left lane line and a right lane line of a lane where a vehicle is located; extracting lane edge target points from the lane line image; conducting perspective transformation on the lane edge target points; and by means of the Hough transformation, detecting a straight line formed by the lane edge target points as a lane line. According to the lane line detection method of the present invention, before the Hough transformation is conducted, the perspective transformation is conducted on the extracted lane edge target points, so that the Hough transformation calculation amount can be reduced, the system reaction time can be reduced, and the travel security of vehicles can be enhanced. In addition, also provided are a lane line detection system, and a method and system for lane deviation prewarning.

A central position of each character participating in a battle is obtained as a fixation point, and multiple temporary points are set on a hemisphere around the fixation point. Each temporary viewpoint position is set on a straight line connecting the fixation point to each temporary point. Each temporary viewpoint position is a position from which all characters can be projected on a virtual screen when perspective transformation is executed at a predetermined visual angle and a distance to the fixation point is shortest. Points are given to each temporary viewpoint position based on the result of a distance evaluation, an angle evaluation, an overlap evaluation, and a height difference evaluation. The top scoring temporary viewpoint positions are selected as positions where the viewpoint of a virtual camera should be moved.

The invention discloses a multi-lane-line detection method. Color images of continuous frames are obtained, graying processing is carried out on a current frame of image, and inverse perspective transformation is carried out on a grey-scale map based on a camera parameter and a set grid map size, thereby obtaining a grid map of a road part; according to priori information of a control point, region division is carried out on the grid map by using a Thiessen polygon and image binaryzation processing is carried out on the divided regions; region grouping is carried out based on priori information of lane lines and pixel coordinates of non-zero pixel values is recorded; curve fitting is carried out on a grouped point set by using an improved RANSAC algorithm, a current image control point is predicted by combining a lane line equation and a particle filter algorithm, a final lane line equation is calculated and e perspective transformation is carried out on the lane line equation, and thus a lane line equation in an original image is obtained. With the method, lane line detection precision and robustness are high; and multi-lane-line detection can be completed simultaneously.

The invention discloses an on-line calibration system for an external parameter of a monocular camera. The system comprises a straight line detection module, a lane extraction module, a fast-perspective-removal (FPR) module, an optimal value iterative module, and an external parameter output module. The straight line detection module obtains an effective straight line position of an inputted infrared image; the lane extraction module carries out effective straight line searching and determination at a possibly occurring lane position; the FPR module carried out perspective transformation removing on a lane based on an imaging principle; the optimal value iterative module carries out loop iteration to enable a fitness function to be optimized, thereby obtaining an angle with the smallest error relative to a practical value; and the external parameter output module is used for outputting the angle obtained by optimal iteration. No calibration object is needed; the influence on the calibration precision by any external factor can be reduced; dependence on the straight line information of the lane is only needed during the calculation process; the flexibility is high; and on-line calibration on the camera can be realized. Moreover, no control by an external device and no strict experiment condition are needed during the calculation process. The system and method are simple and convenient to realize.

In the present invention, whether three-dimensional measurement or checking processing with a model is properly performed by setting information and recognition processing result can easily be confirmed. After setting processing is performed to a three-dimensional visual sensor including a stereo camera, a real workpiece is imaged, the three-dimensional measurement is performed to an edge included in a produced stereo image, and restored three-dimensional information is checked with a three-dimensional model to compute a position of the workpiece and a rotation angle for an attitude indicated by the three-dimensional model. Thereafter, perspective transformation of the three-dimensional information on the edge obtained through measurement processing and the three-dimensional model to which coordinate transformation is already performed based on recognition result is performed into a coordinate system of a camera that performs the imaging, and projection images are displayed while being able to be checked with each other.

Provided herein is a device and a method for recognizing an obstacle and a parking slot to support an unmanned autonomous parking function. The device includes a motion measurement unit measuring a vehicle motion using an in-vehicle sensor, an inverse perspective transform unit performing inverse perspective transformation of an image, which is obtained using a wide-angle camera, to obtain an inverse perspective image, and an obstacle detection unit detecting the obstacle using the inverse perspective image.

The invention discloses a reading method of a pointer type meter based on depth learning, belonging to the field of depth learning and computer vision. The method of the invention utilizes Mask- RCNNobject detection and instance segmentation algorithm to divide the dial and pointer images firstly, then correcting the dial by perspective transformation, Then using PCA (Principal Component Analysis) algorithm to fit the segmented instrument pointer, Then judging the direction of the pointer according to the center coordinate of the smallest oblique circumscribed rectangle of the pointer and thecenter coordinate of the dial; Finally, calculating the pointer reading according to the slope and direction of the pointer by using the angle method. As that method of the invention can accurately classify the type of the instrument, high precision pixel level segmentation of the pointer and dial, under non-uniform illumination conditions, different scales still have good robustness, to solve the traditional pointer instrument recognition field of the dial and pointer positioning difficulties, uneven light, mirror reflection, blurred pictures caused by low recognition accuracy.

The invention discloses a parking space detection method based on a side view camera. Through the segmentation of a passable region in a look-around camera scene, in a ground coordinate system, the actual size of a planar region is calculated through inverse perspective transform, multiple frames of image sequence image data are used to confirm the region with the combination of a vehicle movement state, the position information of an actual parking space region is optimized, and finally the effective parking space detection is realized. According to the method, the reliability of parking space detection can be improved, and the parking accuracy can be improved.

This invention provides a correcting device and a correcting method for perspective transformation of document images. The correcting device comprises a horizontal vanishing point determining unit, for detecting a horizontal vanishing point of the perspective transformed document image; a vertical vanishing point determining unit, for detecting a vertical vanishing point of the perspective transformed document image; and a perspective transformation correcting and converting unit, for correcting the perspective transformed document image; wherein the horizontal vanishing point determining unit comprises a direct horizontal line segment detecting unit, an indirect horizontal line segment detecting unit and a horizontal vanishing point detecting unit, and wherein the horizontal vanishing point detecting unit detects a horizontal vanishing point in accordance with a direct horizontal line segment detected by the direct horizontal line segment detecting unit and an indirect horizontal line segment detected by the indirect horizontal line segment detecting unit.