117results about How to "Reduce matching error" patented technology

A method and a system for processing a semiconductor device intended to improve the overlay accuracy of a semiconductor device product, particularly in its device area, in carrying out the mix-and-match exposure process are designed to calculate the difference of exposure distortions between two layers in the device area and the difference of exposure distortions between the two layers at the overlay measurement mark position from data of exposure field distortions of two exposure tools used for the mix-and-match exposure process and data of device area and overlay measurement mark position of the product, calculate a modification value which relates both differences to each other, calculate a first exposure condition correction value from the measurement result of overlay, and carry out the exposure process based on a second exposure condition correction value which is evaluated by modifying the first exposure condition correction value with the modification value.

The embodiment of the invention discloses a floating car track map matching method and a floating car track map matching device, and relates to the field of an intelligent traffic system. The matching error of the floating car track on a map can be reduced, and the matching accuracy can be improved. The method comprises the following steps that firstly, at least one road respectively corresponding to each track point in a track point sequence is obtained, then, the road matching probability between the track points and each road is calculated according to the distance from the track point to each road in the at least one road, next, the road transfer probability from each road corresponding to the track point to each road corresponding to a next track point is calculated, in addition, a plurality of floating car tracks corresponding to the track point sequence are traversed according to the at least one road respectively corresponding to each track point, and finally, the route tracks of a floating car matched on the map are obtained from the floating car tracks according to the road matching probability and the road transfer probability.

The invention discloses a method for suppressing multiple waves based on L1 norm multichannel matched filtering. The method comprises the steps that a predicted multiple wave channel, a Hilbert transformation channel of the multiple waves, a high-frequency reconstruction channel and translation channels of above three kinds of data are used as multiple channels to be input, so that fitting is conducted with the multiple waves in an original seismic record, and a modulation frequency expanding technology is introduced to reconstruct high-frequency components of the predicted multiple waves. According to the method, a GPU acceleration technology is introduced into an algorithm, GPU/CPU collaborative and parallel computing is used for improving the computational efficiency, and a convergent self-adapting filter is acquired quickly. The method is applied to multiple wave suppressing of a theoretical model and actual data, a test result shows that compared with a conventional L2 form matched filtering method, a high-precision multiple wave suppressing result can be obtained, and meanwhile the energy of effective waves is well protected.

The invention relates to a phase unwrapping method based on colored phase shift stripe secondary encoding. The method comprises the following steps of: firstly generating a colored phase shift stripe, projecting the colored phase shift stripe onto the surface of a measured object, collecting a deformation phase shift stripe graph, transmitting deformation phase shift stripe image data to a computer, utilizing Matlab software to divide the deformation phase shift stripe graph into three phase shift gray level images, solving the phase distribution of the deformation phase shift stripe graph, further solving the wrapping phase of the object, then utilizing the color and luminous intensity alternating rule to realize a first time of colored encoding, carrying out secondary encoding on the basis of the first time of encoding in the sequence of red, green and blue, stipulating that three colored encoding stripes generated by the colored phase shift stripe is a cycle of a stripe graph, encoding the stripes with a two-digit number, wherein the cycle, in which the deformation stripe falls, is represented by encoding of respective regions, and adopting a method of respectively counting the numbers of stripes of each color to determine the cycle number of the stripe in the graph and solve the true phase, thereby reducing the data matching error and promoting the unwrapping speed.

The invention discloses a monocular vision-based 2D plane grabbing method for a mobile mechanical arm. The method comprises the following steps of: calibrating internal parameters of a camera; calibrating a hand-eye relationship; establishing a standard photographing pose and a standard grabbing pose of the mechanical arm; identifying a marker by using a shape-based template matching method; solving the rotation angle of the AGV through a visual servo principle; and calculating a new grabbing pose of the mechanical arm after the AGV moves. The device of the system comprises an AGV mobile robot, a six-degree-of-freedom mechanical arm installed on the AGV, a monocular camera and a clamping jaw which are installed at the tail end of the mechanical arm, a to-be-grabbed object with a fixed position, and a marker arranged beside the to-be-grabbed object. The grabbing pose is calculated by recognizing and positioning the marker, the recognition precision is high, the stability is high, and the method can be suitable for workpieces of different shapes and textures; and meanwhile, a three-dimensional scene is degraded into a two-dimensional scene, so that the complexity and the calculation process of the problem are simplified.

The invention relates to a method for restoring 3D geometric information from images. The method comprises the following steps: a large amount of scene images and the corresponding depth maps thereof are obtained and used as a training library; the images are divided into image blocks, the probabilities thereof corresponding to different depths are described by the statistical learning method to obtain different parameter values; a graph structure including an image block layer, a pixel layer, and the edges connecting the pixels and the corresponding image blocks is constructed for the image pairs used for the scene reconstruction; the energy term of the graph structure is defined by the parameters obtained from the statistical learning; the energy term is taken as a constraint and added to an image describing parallax relation among pixels based on the conventional stereo matching; and a parallax image of the two images is solved.

The invention discloses an optical motion capture data processing method based on a module piecewise linear model. Based on overall information of optical human motion capture scattered data, a data processing algorithm based on the module piecewise linear model is provided. By using the module piecewise linear model to generalize the change characteristics of different modules, the method determines a matching priority and an intra-segment fitting function of module data, effectively performs overall layered prediction and tracking for each three-dimensional motion data module, performs de-noising processing for the noise data based on the module, and provides an interpolation fitting algorithm for the missing motion data based on segmenting Newton to perform reasonable supply. The optimized method does not need manual intervention during processing, and can meet real-time requirement.

The invention belongs to the technical field of the mobile robot visual positioning, and particularly relates to a binocular full-view visual robot self-positioning method based on an SURF algorithm. The method comprises the following steps: (1) setting an artificial road sign as prior position information; (2) matching the features by the improved MDGHM-SURF; (3) eliminating the error matching by a matching point gravity center iterative algorithm; (4) ranging by the three-dimensional backlight path imaging of a vertical-base-line binocular full-view system; (5) positioning by the full-view triangle; and (6) forecasting a detection window to accelerate the positioning speed. The method is capable of, in allusion to the indoor self-positioning problem of the mobile robot, through setting the artificial road sign as the prior position information, executing the feature point rapid detection by using the improved MDGHM-SURF algorithm, executing the road sign identification positioning through the feature matching, reducing the matching error by the matching point gravity center iterative algorithm, improving the positioning accuracy of the road sign center, forecasting the road sign detection area by the robot motion state, and improving the positioning rapidity in the movement.

The invention discloses a vehicle skylight frame and a manufacturing method thereof. The vehicle skylight frame comprises a front cross bar, a pair of curved bars, and a pari of straight bars vertically connected with the two ends of the front cross bar through the curved bars, wherein, the front cross bar, the pair of curved bars, and the straight bars form an integral-forming U-shaped structure; each bar comprises a slide rail slot; and the outer side of the slide rail slot on the straight bar extends outwards by utilizing the slot wall of the slide rail slot, and forms a water drainage slot with the same depth as that of the slide rail slot. The manufacturing method of the vehicle skylight frame comprises a blanking step, a punching step, a cutting step, a stretch bending step, a shaping step and a checking step. The vehicle skylight frame and the manufacturing method thereof adopt an integral-forming structure that no joint exists between the front cross bar and the slide rail, thereby reducing the riveting amount, ensuring the size accuracy and the uniformity, and simultaneously reducing matching errors between the vehicle skylight frame and other part assemblies.

The invention belongs to the technical field of wireless electricity measurement and particularly relates to a distance-Doppler matching method for measuring a radar cross section of a moving target. The invention discloses a high-precision distance-Doppler matching method for measuring the radar cross section (RCS) of the moving target. By utilizing a matching filter with time shifting and a frequency domain subdivision method, the high-precision distance-Doppler matching is realized, and therefore, a high-precision RCS measurement value is obtained. The method has the advantages that compared with a traditional pulse compression method, by utilizing the matching filter with the time shifting, the fluctuation of energy in a distance domain is obviously reduced; compared with a traditional Doppler filter set, by utilizing the frequency domain subdivision method, the fluctuation of energy in a Doppler domain is obviously reduced; and compared with traditional schemes, by utilizing a signal processing method, errors of matching between the distance domain and the Doppler domain are reduced, and therefore, the high-precision RCS measurement value is obtained.

The invention discloses a high-precision compression-configured device of an electro-hydraulic servo valve armature component. The armature component comprises a baffle plate, a tube spring, an armature and a feedback rod which are subsequently arranged in a compression way. The compression-configured device comprises a central component, an adjusting pedestal, a tube spring auxiliary clamping guide mechanism and an armature floatation supporting and positioning mechanism, wherein a longitudinal through hole is formed in the central component; a positioning block used for placing a positioning plate is arranged in the through hole; a compression-configured hole for the tube spring to reach in is formed in the positioning block; the adjusting pedestal is matched with the longitudinal through hole in the central component through screw threads; the top part of the adjusting pedestal pushes against the positioning block to drive the positioning block to move up and down; the tube spring auxiliary clamping guide mechanism comprises at least two transverse through holes which are symmetrical about the longitudinal through hole; a slide positioning baffle block which can extend into the longitudinal through hole to push against the tube spring is glidingly formed in each transverse through hole; the armature floatation supporting and positioning mechanism comprises two elastic component mounting holes symmetrical about the longitudinal through hole left and right; an elastic component and an armature supporting and positioning block which is arranged on the elastic component are arranged in each elastic component mounting hole. The device can assist workers to configure the armature component in a compressing way, so that the compression-configured precision and the production efficiency are greatly improved.

The invention discloses a variable-scale unidirectional gravity sampling vector matching and positioning method based on particle filtering. The variable-scale unidirectional gravity sampling vector matching and positioning method comprises the following steps: step 1, establishing a gravity assisted inertial navigation single-point matching model; step 2, obtaining a matching point based on the particle filtering by utilizing the gravity assisted inertial navigation single-point matching model; step 3, judging whether a gravity assisted inertial navigation single-point matching point is credible or not; and step 4, calculating a final result of a current matching point. According to the variable-scale unidirectional gravity sampling vector matching and positioning method, motorized navigation of a carrier is considered, the disadvantage that an original gravity vector matching algorithm is invalid under the condition of non-linear or non-uniform-speed navigation of the carrier is overcome, and matching errors are reduced.

The embodiment of the present application provides a method and device for generating a high-precision dense depth image, wherein the method comprises the following steps: acquiring a first image, a second image, and a third image; determining a disparity map according to the image; determining preset encoding of a pixel point in a matching window according to the gray value of the to-be-matched pixel point, the gray value of the pixel point in the matching window and a preset encoding rule; determining a matched pixel point from the second image according to the gray value of the to-be-matched pixel point in the first image, the preset encoding and disparity map of the pixel point in the matching window; and further, determining a first depth image. By acquiring and using the preset encoding of the pixel point in the matching window adjacent to the to-be-matched pixel point according to a preset encoding rule, combining the gray value of the to-be-matched pixel, using the disparity map as a constraint, determining the matched pixel point from the second image through matching, and further determining the depth image, the technical problem that the prior method for determining thedepth image is large in error and not high in precision is solved.

A point target movement velocity detection method based on multiple linear moveout scanning, extending and sampling comprises the following steps: (1) establishing a multiple linear moveout scanning detection device; (2) using each linear detector to detect point targets in an extending and sampling manner; (3) processing Nt groups of image data acquired by each linear detector to obtain sub-pixel images; (4) performing non-uniformity correction and sub-pixel match on two sub-pixel images processed by the two adjacent processed linear detectors, then performing difference computation to complete the background subtraction; (5) performing threshold filtering on difference images processed by the two adjacent processed linear detectors, extracting positive and negative point pairs in the difference images by adopting the neighborhood constraint criterion, so as to complete the target detection and extraction of moving points in once scanning process; (6) marking regions of paired positive and negative points extracted in the step (5) respectively, calculating movement velocity and movement directions of the targets according to the position relation of the marked regions of the positive and negative points; (7) averaging the movement velocity and the movement directions obtained in the step (6) to obtain the velocity and the direction of a detected target.

The invention belongs to the technical field of train pantograph visual imaging detection and discloses a pantograph abrasion measuring method and system based on 3D imaging. The method comprises a reference contour acquisition step, a pantograph area positioning step, a positioning discriminator discrimination step, a sliding plate area extraction step, a sliding plate area rotation step, a sliding plate contour coordinate extraction step, a contour coordinate oversampling step and a sliding plate abrasion calculation step. According to the measuring method, a high-speed 3D imaging device installed on the inner wall of a tunnel is used for acquiring image data of a to-be-detected vehicle roof and roof equipment, the real-time position of a pantograph is located through a visual recognition method, and the abrasion condition of a pantograph head is judged by comparing the image data with standard data.

The invention discloses a panorama synthetic method based on image mosaic. Block shooting is carried out on a panorama to obtain block shooting images of the panorama, and the image contact ratio exists between adjacent block shooting images; all block shooting images are converted into corresponding two-dimensional array matrixes according to pixels, coordinates of all the pixels in the block shooting images are expressed by means of a homogeneous coordinate system, and an algorithm based on image gray level similarity and maximum likelihood estimation is adopted to calculate a matrix M; the quadratic sum of image gray level values of minimized adjacent elements is adopted to estimate unknown variables in the matrix M; corresponding nonlinear least squares are solved by adopting an LM iterative algorithm. The algorithm based on image gray level similarity and maximum likelihood estimation is adopted in the method to calculate the matrix M and guarantees optimality of the obtained solution, and feature points easy to identify are not needed to be provided. The calculated amount is relatively less, the matching error is less, and exact match can be achieved.

The invention is applicable to the technical field of mobile robot navigation, and provides a map construction method and device and terminal equipment, and the method comprises the steps: controlling a trolley to move in a target range; collecting laser radar data through a laser radar arranged on the trolley; when the Nth frame of laser radar data is collected, creating an Mth subgraph, wherein N and M are positive integers, and the initial values of N and M are both 1; inserting laser radar data with a preset frame number into the Mth sub-image through related scanning matching to obtain M sub-maps; in the process of inserting the laser radar data with the preset frame number into the Mth subgraph, carrying out closed-loop detection, and carrying out initial value optimization on each frame of laser radar data; and constructing a global map based on the target range according to the M sub-maps. According to the invention, the matching error between each frame of laser radar data and the sub-map and the matching error accumulated by the laser radar data and the sub-map in each sub-map can be reduced, so that the finally constructed global map is accurate and reliable.

The invention relates to a method for precise registration and coincidence of a preoperative CT or nuclear magnetic image and a corresponding focus in an operation. The method is used for assisting planning and implementation of a clinical operation. The method comprises the following steps: acquiring a point cloud of a preoperative CT image or a nuclear magnetic image as a target point cloud; acquiring a point cloud of a focus area of a patient in an operation as an integral point cloud; acquiring a feature region overlapped with the target point cloud from the overall point cloud; adopting a super-4PCS algorithm to carry out registration on the target point cloud and the overall point cloud with the overlapped areas, and obtaining point cloud coordinates of the focus area of the patient in the operation, wherein the point cloud coordinates are space coordinates when the point cloud of the focus area of the patient in the operation and the point cloud of the preoperative CT image or the nuclear magnetic image are registered and coincide. According to the method, the registration precision problem of the three-dimensional model of the non-rigid organs and tissues of the patient and the focus area of the patient in the real operation environment is mainly solved, a doctor can quickly and accurately find the organ corresponding to the three-dimensional model of the patient during the operation, and the three-dimensional model is accurately matched with the organ.

The invention provides a PCB defect positioning method based on multi-template matching. By dividing a template image and according to the characteristics of divided small template images, relative matching templates are arranged, the error in traditional template matching can be greatly reduced through the PCB defect positioning method, the accumulative error caused by actual image deformation can be further reduced through multi-template matching, border-crossing judgment and automatic pixel filling processing are conducted on a matched area after template matching, and the problem that after border-crossing, the defect detection deviation occurs is also effectively solved through the PCB defect positioning method; and through the PCB defect positioning method, the positioning precisionin the PCB detection early stage can be improved, the accumulative deviation between an actual image and a template image is reduced, the detection efficiency and the detection precision are also effectively improved, and a series of detection requirements for high-precision positioning, matching and the like of the PCB detection process are achieved really.

The invention provides a method for enhancing a moving target through multi-linear-array time difference scanning expansion sampling. The method includes the steps that (1) a multi-linear-array time difference scanning expansion sampling detection device is constructed, and Nt detection arrays in each linear array detector in the multi-linear-array time difference scanning expansion sampling detection device are made to conduct imaging at the same time to obtain Nt groups of image data; (2) the Nt groups of image data of each linear array detector are processed to form a frame detection image; (3) after the preset sampling frequency is completed in the scanning direction, the frame detection images obtained by the linear array detectors correspondingly are spliced according to time to obtain two sub-pixel images; (4) non-uniformity correction is conducted on the two sub-pixel images respectively; (5) the sub-pixel image corresponding to the linear array detector which conducts imaging firstly is used as a standard, the sub-pixel image corresponding to the linear array detector which conducts imaging later is moved forwards in the scanning direction; (6) after the two sub-pixel images are matched, difference calculation is conducted to obtain a residual image; (7) enhancement processing is conducted on the residual image after filtering is conducted on the residual image.

The invention belongs to the field of image processing and computer vision, and discloses a depth estimation acceleration method of a multiband stereo camera. In a depth estimation process and a binocular stereo matching process of each waveband, through compressing the matching image, on one hand, a parallel equipotential error brought by binocular image correction is counteracted, so that matching is more accurate, and on the other hand, the calculation overhead is reduced. Besides, before cost aggregation, sparse matching is carried out by carrying out transverse compression on the cost graph, so that the calculation overhead is reduced again. And the disparity maps obtained in different modes are fused to obtain all-weather, more complete and more accurate depth information.