375 results about "Scale transformation" patented technology

The invention relates to a remote headnote moving and scaling synchronization method and a remote headnote moving and scaling synchronization system. The remote headnote moving and scaling synchronization method comprises the following steps: acquiring a coordinate set of a current screen image drawn and stored on a drawing canvas; when monitoring a screen touch event, acquiring a translational displacement and / or a scaling ratio of the screen touch event; according to the translational displacement and / or the scaling ratio, translating and / or scaling the coordinate set to obtain a translated and / or scaled transformation matrix; according to the transformation matrix, transforming an original coordinate system of the current screen image to obtain a transformed coordinate system; according to the transformed coordinate system, redrawing the current screen image to obtain a redrawn screen image; according to the transformed coordinate system, calculating to obtain coordinate information and a scaling multiple of the redrawn screen image; sending the coordinate information and the scaling multiple to an interactive smart tablet; and executing corresponding operations by the interactive smart tablet according to the coordinate information and the scaling multiple.

Techniques for registration of multiple measurement modes of a body include receiving first and second data from different modes. Each includes measured values with coordinate values. For two mechanically aligned modes, any nonrigid registration is performed. For some modes, the nonrigid registration includes a coarse transformation and multiple fine scale transformations. The coarse transformation maximizes a coarse similarity measure. The second data is subdivided into contiguous subregions. Fine transformations are determined between the subregions and corresponding portions of the first data to maximize a fine similarity measure. Subdividing and determining fine transformations repeats until stop conditions are satisfied. Transformations between the last divided subregions are interpolated. Any of the fine similarity measure, a search region, interpolation method, sub-division location, and the use of rigid or non-rigid fine transformations are adaptive to properties of the first or second data so that the registration is automatic without human intervention.

The invention discloses a visual detection method for the transparent paper defect of the outer package of a strip cigarette, which aims to improve the efficiency of visual detection for the transparent paper defect of the outer package of the strip carton The method comprises the following steps: performing gray scale transformation on the shot digital image of the strip cigarette, filtering to remove noise and performing pretreatment on the image; performing edge detection on the pretreated image and extracting an edge image; performing binaryzation treatment on the edge image, performing gray value statistics in the horizontal and vertical directions by using a projection method, and extracting the boundary of the strip carton image to realize separation of the strip cigarette image; calculating similarity between the separated strip cigarette image and a template image, and setting a judgment threshold; and finally comparing the similarity value and the threshold to determine whether the strip cigarette image is qualified, extracting a pixel image of the defect part of the unqualified strip cigarette image by using a difference image method, and counting the pixel value which is taken as the pixel area of the defect part.

The invention discloses a method for segmenting HMT images which is based on the nonsubsampled Contourlet transformation. The method mainly solves the problem that the prior segmentation method has poor area consistency and edge preservation, and comprises the following steps: (1) performing the nonsubsampled Contourlet transformation to images to be segmented and training images of all categories to obtain multi-scale transformation coefficients; (2) according to the nonsubsampled Contourlet transformation coefficients of the training images and the hidden markov tree which represents the one-to-one father and son state relationship, reckoning the model parameters; (3) calculating the corresponding likelihood values of the images to be segmented in all scale coefficient subbands, and classifying by examining possibility after integrating a labeled tree with a multi-scale likelihood function to obtain the maximum multi-scale; (4) updating category labels for each scale based on the context information context-5 model; and (5) with the consideration of the markov random field model and the information about correlation between two adjacent pixel spaces in the images to be segmented, updating the category labels to obtain the final segmentation results. The invention has the advantages of good area consistency and edge preservation, and can be applied to the segmentation for synthesizing grainy images.

The invention belongs to the field of agricultural remote sensing, and relates to a crop yield estimation method based on scale transformation and data assimilation and application of the crop yield estimation method to the guiding of crop production. The method comprises the following concrete steps that: 1, parameters are collected for completing spatialization of a WOFOST crop model; 2, a crop type distribution map and a purity percentage map of the crops to be tested are obtained; 3, TMLAI in a range of 30m is obtained; 4, a secondary scale conversion model is built, and a time sequence scale regulating LAI is generated; 5, the crop model error and the remote sensing observation error in a key phonological period of the crops to be tested are obtained; 6, a four-dimensional variation cost function is built, and an optimized crop model parameter is obtained; and 7, the per unit yield of the crops to be tested in a county region is output. The method provided by the invention has the advantages that the problem of scale mismatching between a remote sensing observation picture element and a crop model simulation unit is solved; the precision of a data assimilation model is improved; and the method is suitable for the crop yield estimation in the county region scale, and is particularly suitable for the winter wheat yield estimation in the county region scale.

The invention discloses a human body pose estimation method. The human body pose estimation method is suitable for marking human body key points in the video to be processed so as to extract the humanbody pose information, and is suitable for application in a mobile terminal. The human body pose estimation method includes the steps: acquiring the video to be processed, and determining whether thekey point position of the previous image frame of the current image frame exists; if the key point position of the previous image frame does not exist, detecting the human body position in the current image frame, and generating a target frame containing the human body position; if the key point position of the previous image frame exists, according to the human body key point position marked inthe previous image frame, generating a target frame corresponding to the current image frame; performing scale transformation on the generated target frame to generate a human body image with the preset size; inputting the human body image with the preset size into a preset pose estimation model so as to output the image on which the human body key point position is marked; and according to the marked key point position, estimating the human body pose. The invention also discloses a corresponding mobile terminal.

The invention discloses a vehicle detecting method based on Gauss difference multi-scale edge fusion. The method includes the steps that Gauss scale transformation is performed on images to obtain four Gauss images in adjacent scales; according to the four Gauss images in the adjacent scales, the difference operation is performed between the images in the adjacent scales to obtain three Gauss difference images different in scale, edge detection is performed on the obtained three Gauss difference images through a Sobel operator, then edge fusion with the scale upward searching is performed to remove a lot of background edges while edge information of a vehicle is obtained as much as possible, and expansion, closed operation, hole filling and other series of morphological operation are performed on the fused edge images to obtain a connected domain image representing the vehicle; an outside rectangle of the position where the vehicle is located is determined in the original image according to the position information of a connected domain to detect the vehicle. The images in multiple scales are processed, so that algorithm complexity is reduced, operation amount is reduced, efficiency of vehicle detection is effectively improved, and a good detection result is obtained.

The invention provides a high-efficiency registration method aimed at a CT and optical scanning tooth model, and belongs to the field of computer oral cavity recovery. The method comprises the steps of (1) data preparation in which 3D visualization is carried out on CT data of the oral cavity to obtain a CT reconstructed tooth image, visualization is carried out on optical scanning data of the tooth crown and gum part to obtain an optical scanning tooth crown image, and the CT reconstructed tooth image and the optical scanning tooth crown image are arranged in the unified world coordinate system; (2) initial registration in which characteristic points are selected manually to process the CT reconstructed tooth image and the optical scanning tooth crown image and further to obtain an initial registration model; and (3) accurate registration in which characteristic points are selected automatically to process the initial registration model and further to obtain an accurate registration model. Problems in registration data with scale transformation and different resolution are solved, and a registration result is high in convenience speed and low in registration error.

The invention discloses methods for calibrating machine tool follow-up laser scanning coordinates on the basis of space standard balls, belongs to the field of vision measurement, and relates to a method for calibrating machine tool follow-up laser scanning coordinates on the basis of space standard balls. The method includes providing third-dimension data for linear laser measurement sensors by movement modules of machine tools; scanning, measuring and fitting circle centers of the space standard balls with known radiuses and unifying data of coordinate systems of the machine tools and linear laser measurement coordinate systems; calibrating default axes of the measurement coordinate systems according to movement of the machine tools and movement of laser planes; calibrating measurement axes; integrally calibrating three-dimensional movement to obtain scale transformation matrixes of the coordinate systems of the machine tools and the measurement coordinate systems. The method has the advantages that the shortcoming that only two-dimensional data can be measured by the traditional laser measurement sensors is overcome, the measurement application ranges of the linear laser measurement sensors can be effectively expanded, and the method has a broad application prospect.

The invention discloses a method for copying joint waviness by an artificial rock material and an experimental method. The method includes the following steps: firstly, measuring the joint surface contour line waviness by a three-dimensional laser scanning method; refining a graph and removing noise points from the graph by contour tracking, performing image scaling transformation and coordinate transformation on a joint contour line, and obtaining coordinate values of points on the contour line by an interpolation method; reading joint surface contour line data, digitalizing the joint surface contour line waviness, introducing the digitalized joint surface contour line waviness into a spark cutting control machine, processing a steel block by 2Cr13 alloy steel and making mortar joints; and laying joint mud or mineral clay with different thicknesses and water contents between the mortar joints, and carrying out a mechanical experiment of the joints. The method disclosed by the invention can be used for truly reflecting the physical states of filler rock masses containing joints, the influence of the joints to the strength and deformation properties of the rock masses can be researched by subsequent experiments, and the method is suitable for a variety of rock masses ranging from high-quality ones to broken ones.

Techniques for registration of multiple measurement modes of a body include receiving first and second data from different modes. Each includes measured values with coordinate values. For two mechanically aligned modes, any nonrigid registration is performed. For some modes, the nonrigid registration includes a coarse transformation and multiple fine scale transformations. The coarse transformation maximizes a coarse similarity measure. The second data is subdivided into contiguous subregions. Fine transformations are determined between the subregions and corresponding portions of the first data to maximize a fine similarity measure. Subdividing and determining fine transformations repeats until stop conditions are satisfied. Transformations between the last divided subregions are interpolated. Any of the fine similarity measure, a search region, interpolation method, sub-division location, and the use of rigid or non-rigid fine transformations are adaptive to properties of the first or second data so that the registration is automatic without human intervention.

The invention provides an electrocardiosignal construction method based on pulse wave signals. The method comprises the following steps: performing normalization processing of pulse wave signals and electrocardiosignals, which are segmented periodically, by means of scale transformation, fitting the pulse wave signals and the electrocardiosignals by adopting a cubic b-spline curve, and extracting characteristic parameters, transforming the characteristic parameters by a principal component analysis method to obtain the principal components of the signals; performing sample training of the principal components of the pulse wave signals and the electrocardiosignals by adopting a BP neural network to obtain model parameters, and constructing a forecasting model; and constructing electrocardiosignal parameters through the model by utilizing the pulse wave signals and the network model parameters. Experiments show that the method provided by the invention has favorable algorithm performance and constructed electrocardiosignals have small distortion. According to the method, the electrocardiosignals are constructed according to pulse waves through a portable waist type pulse detector, the use of chest leads are avoided, and the improvement in portability of an electrocardiograph is facilitated.

The invention discloses a vehicle speed tracking fuzzy control method of a vehicle robot driver, comprising the following steps: (1) determining an input variable and an output variable of a fuzzy controller of the vehicle robot driver, and carrying out scale transformation on an actual input variable to lead the actual input variable to be transformed in a domain range; (2) using the fuzzy controller of the vehicle robot driver to carry out fuzzification on the actual input variable after scale transformation to obtain a fuzzy value, and using the fuzzy value to establish an input and outputvariable membership function; (3) establishing a fuzzy control rule of the vehicle robot driver based on a learning method of a driving experience knowledge base and measuring data; (4) using a fuzzyreasoning and a centroid defuzzification method to determine a fuzzy control table of the vehicle robot driver based on the input and output variable membership function and the fuzzy control rule; and (5) querying the fuzzy control table of the vehicle robot driver to perform online real-time control on the vehicle robot driver. The invention features good instantaneity, strong anti-interferenceability and fine vehicle speed tracking control precision.

The invention discloses a target scale adaptive tracking method based on OpenCV. The method comprises the following steps of: positioning and tracking a target by utilizing a mean shift algorithm; extracting target features by using SIFT (Scale Invariant Feature Transform) every 6 frames; carrying out feature matching on the template image and the tracking real-time image, and determining a mapping relation between matching points through affine transformation to obtain a scale transformation factor, thereby updating the size of a candidate region target template in the real-time image, and solving the problem that tracking is easy to lose or track mistakenly because the scale of a moving target is continuously increased.

The invention provides a vehicle character recognition method based on a CNN convolutional neural network, and the method comprises the following steps: S1, building an HAAR cascade classifier, locking the position of a license plate according to the HAAR cascade classifier, and extracting an image of the license plate; S2, carrying out gray scale transformation on the image of the license plate to form a gray scale image; S3, performing image binarization operation on the grayscale image; S4, after the gray scale image is binarized, performing image corrosion and image expansion operation toobtain a final image; and S5, finding the information of the characters in the license plate by the final image through the minimum image contour, transmitting the characters into the established CNN,and carrying out recognition operation on the characters to finally obtain the license plate information. The efficiency of character detection is improved, and the license plate recognition time isshortened.

The invention discloses a method and a device for generating a depth map, which belongs to the technical field of computer vision. The method comprises the following steps: performing gray-scale transformation by using a left view and a right view of an input image pair as a reference image and a target image respectively; computing upper, lower, left and right gradient matrixes of the reference image after the gray-scale transformation; computing a cost function matrix according to parallax, and processing borders; computing an original information storage matrix, and initializing upper, lower, left and right information storage matrixes; performing iterative calculation in a GPU memory by using a CUDA belief propagation algorithm to acquire upper, lower, left and right information storage matrixes after iteration, and computing a belief matrix according to the result of the iteration; and computing the depth map according to the belief matrix. The device comprises a gray-scale transformation module, a gradient computation module, a cost function matrix computation module, an original information storage matrix computation module, an iteration module, a belief matrix computation module and a depth map computation module. The method and the device improve computing efficiency of the depth map and realize high-quality and fast generation of the depth map.

The invention relates to a display brightness adjustment method and device. The display brightness adjustment method comprises the steps of obtaining original gray-scale values of pixel points on a display interface, conducting gray-scale transformation on the pixel points to obtain transformed gray-scale values of the pixel points, and adjusting gray scale of the interface according to the transformed gray-scale values of the pixel points.

Computer-implemented methods and computer program products for automatically transferring expressions between rigs with consistent joint structure, and for automatically transferring skin weights between different skin meshes based on joint positioning. A method is provided for transferring an expression between a source rig and a target rig, where each rig characterizes an animated character, and each rig, in turn, is characterized by a set of joints and a skin mesh having a plurality of vertices, with each vertex characterized by a matrix of weights relating a response of the vertex to movement of associated joints. A set of offsets is calculated of joint positions of a goal expression of the source rig relative to a neutral expression of the source rig. A scaling transformation is then applied to the set of offsets to produce a scaled set of offsets, which are added, in turn, to a neutral expression of the target rig. Methods are also provided for transferring a set of skin weights between a source rig and a target rig.

The invention relates to a system for simulating the regional rainfall process, which comprises a hydro-meteorological database, a rainfall process generation subsystem, a rainfall process control subsystem and a rainfall simulation subsystem, and further comprises a regional climate model database, a station information database, a regional climate model, a station information interpolation device, a scale transformation device, a water spray control device, a water supplying device, a spray device, and a sensor mounted on a regional physical model, wherein spray units are arranged crosswise to form a spray unit array. Through the integration of the hydro-meteorological database, the rainfall process generation subsystem, the rainfall process control system and the rainfall simulation units, the generation and the simulation of the past, present and future rainfall processes in the designated area are achieved. The natural rainfall process can be better simulated, and the system has an important scientific research value in the fields of the watershed runoff generation and flow routing process, the research on hydrology and water resources in the lack-data watershed, the inspection and the optimization of the hydrographic station net arrangement, and the intelligence watershed research, and provides a scientific tool for simulation study and application of the regional hydrology.

The invention relates to a fundus image blood vessel segmentation method based on Frangi enhancement and an attention mechanism UNet, and the method comprises the steps: firstly extracting a green component from an input image, and carrying out the contrast adjustment on the basis of the extracted green component through a contrast-limited histogram equalization method; calculating a Hessian matrix of each pixel point in the image after the contrast ratio is adjusted; constructing a Frangi vascular similarity function by utilizing the characteristic value of the Hessian matrix under the condition of a scale factor, and obtaining the maximum response; respectively subtracting the product of the maximum response value and the enhancement factor factor factor from the pixel values of the RGBthree same channels of each pixel point of the input image; then, carrying out gray scale transformation on the image after frangi enhancement, and carrying out zero mean normalization operation on each pixel value to be between [0, 1]; and finally, inputting the obtained training image blocks and label image blocks into an attention mechanism UNet network for training; and obtaining a segmentation result through testing. According to the invention, the generalization ability of the model is improved.

The invention discloses a video abnormity detection method based on a graph structure under multi-scale transformation. First of all, video abnormity detection is carried out by use of an optical flow feature, and spatial correlation scales of video abnormity association optical flow features in different scenes are different. Under the condition that the spatial correlation structure of the video abnormity association optical flow features is maintained, since the calculation efficiency of video abnormity detection can be effectively improved by reducing the quantity of the optical flow features, it is brought forward that a network graph structure of the optical flow features is constructed based on spatial correlation, by use of multi-scale transformation of the graph structure under correlation constraints, the quantity of the optical flow features in the video abnormity detection is effectively reduced, under the condition that the detection precision is slightly reduced, the calculation efficiency of a video abnormity detection algorithm can be substantially improved, and detection of event abnormities is more rapid and reasonable.

The invention belongs to the technical field of image processing, relating to a characteristic matching method based on bilateral matching and trilateral restraining. The method comprises the following steps of: obtaining initial points with a bilateral matching arithmetic; reorganizing the initial points with trilateral triangulation arithmetic; and using the characteristic matching arithmetic based on the trilateral restraining to achieve a result for improving the correction rate of matching points and increasing the number of the correct matching points. For most geometrical transformation and light transformation, the method comprises rotation transformation, scale transformation, fuzzy transformation, visual angle change, JPEG compression and light intensity change, is more robust, and can obviously improve the correction rate of the matching points and the number of the matching points, thereby having wider application field, being capable of obtaining a better characteristic matching result, and having higher practical applicability.

The invention relates to a remote sensing image thin cloud removal method based on a multi-scale collaborative learning convolutional neural network. According to the method, training data is input into network structures with different scales after being subjected to scale transformation; features under different scales are extracted level by level from coarse granularity to fine granularity sequentially for fusion, so that coarse-to-fine multi-scale learning is realized; and last, a mapping relation between cloudy data and cloudless data is obtained, cloud components are effectively removed,image details are restored, and the purpose of thin cloud removal is achieved. An experiment result indicates that compared with a traditional thin cloud removal method, the method can eliminate a manual trace brought by the traditional method, information of a cloudy region in an image can be accurately restored, the fidelity of a cloudless region can be maintained, the thin cloud removal effectis better, and thin cloud removal precision is higher.

The invention discloses an image retrieval method based on multiple intelligent algorithms and an image fusion technology. According to the method, on the basis of a reinforced learning and genetic algorithm and a clustering algorithm, an intelligent learning framework for active learning is constructed; the image fusion technology and the genetic algorithm are used during relevant feedback, so that an inquiry vector and a similarity matching model are corrected; and the inquiry precision and the inquiry efficiency are improved. The image retrieval method has the advantages of higher inquiry precision and inquiry efficiency and relatively high robustness for translation, rotation and scale transformation; furthermore, after a certain number of inquiries and learning, intelligent retrieval can be realized; and the inquiry precision and the inquiry efficiency are further improved.

The invention discloses a lightweight video interpolation method based on a feature level optical flow, which is used for solving the technical problem that the existing lightweight video interpolation method is poor in practicability. According to the technical scheme, firstly, multi-scale transformation is carried out on two continuous frames of images in a given video, and an optical flow estimation module at the feature level is adopted to calculate a forward optical flow and a reverse optical flow between the two frames under the current scale; warp transformation on the two images is performed in time sequence according to the forward optical flow and the reverse optical flow respectively to obtain two interpolation images; the interpolated images are combined to obtain a four-dimensional tensor, and the tensor is processed by utilizing three-dimensional convolution to obtain an interpolated image under the scale; And weighted average on the images are carried out with differentscales to obtain a final interpolation image. According to the invention, the video interpolation is carried out by using the optical flow of the feature level and the multi-scale fusion technology, and the precision and speed of the video interpolation are improved. A 1.03 MB network model is used for obtaining the average peak signal-to-noise ratio of 32.439 and the structural similarity of 0.886.

An image segmentation method based on a Gaussian smoothing filter is provided. The method comprises the following steps: carrying out a weighted average operation on three components (RGB) by means of different weights according to the importance of the three components and other indexes; adopting histogram equalization to correct a histogram at first, correcting the histogram of an original image and enabling the histogram to be evenly distributed through a gray scale transformation function, and then conducting histogram equalization; conducting Gaussian smoothing filtering on the image, adopting a method utilizing only one global threshold T during a binary process, comparing the gray value of each pixel of the image with the T, and taking a foreground color if the gray value is greater than the T, otherwise, taking a background color; taking a central point P corresponding to the maximum number of pixels within L gray scale ranges as an initial mean value; checking each pixel during the i iteration, and calculating the distance between each pixel and the mean value of each gray scale, and giving each pixel the mean value closest to the class of the pixel; for j=1,2,...1, calculating a new cluster center, and updating a class mean value; and checking all the pixels one by one.

The invention discloses a gesture recognition method based on an improved residual neural network. The method includes the following steps: S1, acquisition of a training sample set; S2, preprocessingon the training sample set, wherein positions of gestures in images are found through algorithms, and cropped images are used as original training data; S3, enhancement of training samples, wherein translation transformation, rotation transformation, mirror-image transformation, scaling transformation and the like are carried out on the collected training samples to enlarge the training sample set; S4, acquisition of a gesture model, wherein a processed training sample set is input into the pretrained residual network to carry out training on network parameters to obtain the gesture recognition model; S5, a step of carrying out processing, which is the same as the step S2, on to-be-recognized gesture images to obtain to-be-recognized gesture data; and S6, a step of inputting the to-be-recognized gesture data into the network, of which training is completed, to obtain a gesture sequence. The method is based on the deep residual network, trains the residual network on the self-collecteddata set, and realizes high-recognition-rate gesture recognition of a third view angle.

The invention provides an icon recognition method and device. The method comprises the following steps: a to-be-recognized image containing a specific icon is acquired; an image processing method is used to process the to-be-recognized image to detect the feature points of the specific icon, and feature vectors of the feature points are extracted to form a unique feature vector group; the generated feature vector group and a feature vector group uniquely corresponding to reference icons preset in a feature database are matched one by one to determine a target icon matched with the specific icon; and the icon information corresponding to the target icon is acquired from a third-party database and is outputted. The method carries out icon recognition based on an SIFT algorithm with scale invariance, the icon recognition precision is enhanced, and robustness of the scheme towards rotation, brightness change and scale transformation happening to the image are enhanced.

The invention discloses a time-variant underwater acoustic channel capacity simulation model. The time variant underwater acoustic channel capacity simulation model comprises a storage module for storing various parameters, a reading module for reading the various parameters stored in the storage module, a large-scale module for obtaining random channel parameters subjected to large-scale transformation through first-order AR transformation, a multipath channel processing module for obtaining path information of multiple paths of the random channel parameters based on a ray tracing model, a screening module for screening out three to ten important paths, a small-scale module for calculating small-scale attenuation coefficients of multiple micro-paths according to small-scale parameters, a Doppler module for processing Doppler parameters to calculate an effective Doppler matrix, a channel transmission function generating module for obtaining a total channel transmission matrix of a multipath time delay acoustic channel according to comprehensively-obtained various parameter, and a channel capacity processing module for conducting processing transformation on channel transmission functions in the total transmission matrix and obtaining pulse response, channel gain and channel capacity of the acoustic channel through calculation.