37 results about "Pyramid algorithm" patented technology

The present invention provides a fast high-precision geometric template matching method enabling rotation and scaling functions. According to the method, based on image edge information, with the sub-pixel edge points of a template image adopted as feature points, the tangent directions of the gradient directions of a target image adopted as feature lines, and based on the local extremum points of a similarity value, the similarity and the candidate positions of the template image and the target image are calculated from fine to rough through adopting a Pyramid algorithm and according to a similarity function; and pixel-level positioning accuracy is obtained at the bottommost layer of a Pyramid, the least squares method is adopted to carry out fin adjustment, so that sub-pixel positioning accuracy, higher-precision angle and size scaling accuracy can be achieved. The method can realize fast, stable and high-precision positioning and identification of target images which are moved, rotated, scaled and is partially shielded, and where the brightness of illumination changes, illumination is uneven, and cluttered background exists. The method can be applied to situations which require machine vision to carry out target positioning and identification.

The invention brings forward a rapid and automatic mosaic technology of aerial video in a search and tracking system. The technology provided by the invention comprises the following steps: two continuous frames of color images in an aerial video sequence are converted to grayscale images; three-valued processing is carried out on the grayscale images by an image three-valued method, edge feature information images are extracted by an edge detection sobel operator and feature points are extracted by an Harris algorithm; feature point matching is completed by an optical flow pyramid algorithm; optimal homography matrix of the color images is calculated by combining a DLT algorithm and a RANSAC algorithm; panoramic image mosaics is completed after image correction of the color images by correction geometric transition matrix; and dynamic regulation of image three-valued upper and lower thresholds and a threshold of the edge detection sobel operator is carried out during the process of panoramic image mosaics. According to the invention, image mosaic accuracy and robustness are raised.

The invention relates to the cross field of geographic information and information technology, in particular to a distributed algorithm for quickly establishing a massive remote sensing image pyramid applied to the construction of National Spatial Data Infrastructure. The method realizes the algorithm for quickly establishing the massive remote sensing image pyramid by means of data distribution storage and parallel computation capabilities based on a PC (Personal Computer) cluster, through the steps of dividing N*M grids, performing pyramid establishment task distribution and scheduling and pyramid establishment and correctly dividing grids and computing grid information, and by means of a communication transfer task between a control node and a computation node; and the method mainly aims to solve the related technical problem of how to divide the N*M grids, etc. The distributed algorithm for quickly establishing the massive remote sensing image pyramid has the following advantages that: based on the data distribution storage and parallel computation capabilities, the speed of generating the massive remote sensing image pyramid is increased; and the problem of black edges causedby image block documents crossing grids is solved; therefore, the purpose of increasing the pyramid establishing speed is achieved and the pyramid establishment accuracy is ensured.

An irregular connectivity mesh representative of a surface having an arbitrary topology is processed using a non-uniform relaxation procedure. The non-uniform relaxation procedure minimizes differences between vectors normal to faces of pairs of triangles having a common edge and located within a designated neighborhood of a given vertex. The relaxation procedure may be used to construct subdivision and pyramid algorithms for performing processing operations such as upsampling, downsampling and filtering on irregular connectivity meshes. The signal processing algorithms may be utilized in applications such as smoothing, enhancement, editing, texture mapping and compression.

The invention relates to a gesture recognition method based on an image space pyramid bag of features. The method comprises the following steps of carrying out gesture division of an image with a gesture; extracting and describing features of the divided gesture image; using extracted features to train a histogram intersection core support vector machine; obtaining a gesture type to which a feature vector of the gesture image belongs according to the support vector machine, in order to achieve gesture recognition. The method combines a space pyramid algorithm and a bag of features algorithm, and describes features of overall quantity and distribution of feature points of the gesture image. By means of the histogram intersection core support vector machine, classification of gesture features is achieved, and gesture recognition is achieved further. In addition, the rate of recognizing gestures of multiple similar types is improved.

The invention discloses an identification method for a tiny leakage signal of an oil transportation pipeline, relating to the field of pipeline signal processing. The identification method comprises the following steps: selecting a db4-10 series wavelet function as a wavelet base for wavelet conversion and determining that the quantity of layers of wavelet decomposition is four; carrying out four-layer discrete wavelet decomposition on an original sound signal by adopting a MALLAT pyramid algorithm; selecting detail signals of a third scale and a fourth scale and carrying out wavelet reconstruction to obtain de-noised signals; uniformly dividing the de-noised signals according to time to obtain divided sound signal segments; carrying out short-time Fourier transform on each divided sound signal segment to obtain a conversion matrix; manufacturing a normalized energy pattern of the sound signals by utilizing the conversion matrix; and judging whether the oil transportation pipeline has tiny leakage or not according to the normalized energy pattern. According to the identification method disclosed by the invention, effective identification of the leakage signal is realized and various requirements in actual application are met.

The invention provides a color image fusion method based on an S-PCNN and a Laplacian pyramid. An effective color image fusion method based on an S-PCNN (simplified pulse coupled neuronal network) and a Laplacian pyramid is brought forward. In HSV color space, after the S-PCNN and local entropy are used for carrying out feature region clustering on an H component, fusion of H components of source images is realized on the basis of a pulse oscillation frequency diagram; the Laplacian pyramid is used for carrying out resolution decomposition on an S component and a V component, and then different fusion rules are used for fusing the S component and the V component; and finally, color space inverse transformation is carried out on the new H component, the S component and the V component, and fusion of an RGB color image is realized. The experimental result shows that the algorithm of the invention is better than other common image fusion algorithm both in the aspect of subjective visual effects and in the aspect of objective evaluation standard.

The invention provides a tropical cyclone intensity objective determination method based on a satellite cloud chart and an RVM. The method is used for constructing a tropical cyclone (TC) intensity objective determination model based on the satellite cloud chart and the RVM (relevance vector machine). The method mainly comprises the following two aspects: 1) carrying out fusion on infrared and water vapor channel cloud charts by utilizing a Laplacian pyramid algorithm, constructing a deviation angle-gradient co-occurrence matrix with the TC center as a reference point, constructing characteristic factors closely related to TC intensity by utilizing a plurality of statistical parameters in the co-occurrence matrix and information of TC kernel scale and center latitude and the like, and establishing the TC intensity objective determination model by utilizing the RVM; and 2) based on the fused satellite cloud chart, and with each point as the reference point in sequence, constructing a deviation angle-gradient co-occurrence matrix and calculating a minimum value, a median value and a mean value of a co-occurrence matrix statistical parameter array. The method constructs the characteristic factors closely related to TC intensity by utilizing the plurality of statistical parameters of the co-occurrence matrix parameter array and information of TC kernel scale and center latitude and the like, and establishes the TC intensity objective determination model by utilizing the RVM.

The present invention belongs to the digital image processing technical field and discloses a Harris corner detection image pyramid palm print ROI (region of interest) recognition method. The method includes the following steps that: an entered palm print ROI image and a template palm print ROI image are inputted, MFRAT filtering is performed on the two images; the Harris corner detection algorithm is used to extract corners distributed on palm print main lines and wrinkles from the filtered template palm print ROI image with only one corner reserved in each small block region; the image pyramid algorithm is adopted to obtain corners distributed on the filtered entered palm print image according to the corners on the filtered template palm print ROI image; and the average BLPOC (Band-Limited Phase-Only Correlation) algorithm is adopted to obtain a final matching score according to known corner distribution; and a fixed threshold value is set according to an experiment, if the matching score is larger than the threshold value, a palm print is a true match, otherwise, the palm print is a false match. With the Harris corner detection pyramid palm print ROI (region of interest) recognition method of the invention adopted, palm print image ROI matching can be performed accurately.

The invention discloses a solar cell panel crack detection method comprising collecting a solar cell panel image; dividing the solar cell panel image into a plurality of singlechips by horizontal vertical projection, and according to the horizontal projection of singlechip broken gates, cutting the singlechips into multiple blocks; based on Laplacian Pyramid, decomposing the blocks, and using canny algorithm for edge detection of the decomposed blocks to obtain the characteristics of a solar cell panel crack; performing denoising processing on the characteristics of the solar cell panel crack;determining the position of the solar cell panel crack based on the denoised characteristics of the solar cell panel crack. The result of the edge detection based on the Laplacian pyramid algorithm is more complete and smooth, influence of partial noises can be effectively removed, and the accuracy of detection is improved.

The invention discloses a fire-fighting unmanned aerial vehicle image fusion method based on visible light and infrared thermal imaging. The method comprises the steps: using a visible light camera and an infrared thermal imaging camera of a fire-fighting unmanned aerial vehicle to capture a visible light image and an infrared thermal image respectively; converting the two pictures into an image with the size of 1280 * 720 through a size function in the OpenCV; extracting ROI vertex coordinates in the infrared thermogram and displaying the ROI vertex coordinates in the visible light image by using a red rectangular frame; fusing an image in the edge obtained in the infrared thermogram and an image of a coordinate corresponding to the visible light by using a Laplace pyramid algorithm to obtain fire point information; and after the fire point is detected, calculating the three-dimensional position of the target according to the visual information, and accurately estimating the three-dimensional position information of the fire point under the coordinate system of the fire-fighting unmanned aerial vehicle by calculating the relative height and yaw angle between the unmanned aerial vehicle and the target. The method has the advantages of being simple in structure, convenient to operate and easy to implement.

The invention discloses a method for splicing double fisheye images and fusing a panoramic image seam. For two fish-eye expanded images which go through characteristic registering and geometric perspective transformation correction and are superposed in overlapped areas, the method for splicing the double fisheye images and fusing the panoramic seam firstly uses an overlapped area of the two images to calculate a brightness color adjustment function between the two images and adjusts the brightness colors of the two images to be the same, and then utilizes a pyramid algorithm to perform fusion on the seam through reasonably setting a weight value at the seam between the two images. The method for splicing double fisheye images and fusing panoramic image seams solves a problem the brightness of two spliced images is not uniform and the colors are different, removes the seam at the splicing part between the two images and makes the two images perfectly splice.

The invention discloses a multi-scale X-ray image enhancement method based on the characteristics of the human visual system. , and perform hard copy simulation; grayscale processing, channel decomposition to obtain cone images and rod images; use Laplacian pyramid algorithm to perform multi-scale decomposition of images; perform brightness and contrast adaptation; Threshold processing; images are merged respectively, and inverse sensing and inverse gain processing are performed on the merged images to realize the image reconstruction of the Laplacian pyramid, and the reconstructed image is obtained; the reconstructed image is normalized, and the normalized image is obtained. Perform color space conversion and gamma correction to obtain the corrected RGB color space image and output. The invention can effectively enhance the contrast and details of the X-ray image, and can better reduce the influence of noise on the X-ray image.

The invention discloses a shearing wave image denoising method based on a cutoff window, which mainly solves the problem of large time complexity of a traditional shearing wave based on a first basis function algorithm and the problems of inflexible frequency band division and poor denoising effect of the traditional shearing wave based on a traditional Laplace pyramid algorithm, and comprises the following steps of: carrying out discrete Fourier transformation on a noisy image and the construction on a rectangular cutoff window or a round cutoff window; obtaining an approximate image through a rectangular central cutoff window or a round central cutoff window, carrying out direction division on the frequency bands of the rectangular cutoff window or the round cutoff window acting on the frequency domain of the noisy image by the shearing wave to obtain a coefficient matrix; afterwards carrying out hard threshold denoising on the coefficient matrix; and then reconstructing the denoised coefficient matrix and the approximate image by combining the rectangular cutoff window or the round cutoff window with the shearing wave to obtain the denoised image. The cutoff window constructed by the invention can obtain more ideal effect on image denoising and can be used for image analysis.

The invention provides a large-scale difference image feature point matching method based on a deformation space pyramid. The method comprises: taking two images with large resolution scale differences as objects, firstly, adopting the Dense SIFT algorithm to extract feature points and their descriptions from a left-to-right, top-to-bottom overlap on a fixed-size grid or bounding box by using a fixed step length, then matching the feature points by using a deformation space pyramid algorithm, and finally eliminating mismatching points by using RANSAC to obtain a final matching result. The algorithm matching effect is very good; The method can provide multi-source data cooperative technical support for land resource dynamic remote sensing monitoring of land resource management departments and production units, can be applied to the aspects of geological survey, mine monitoring, agriculture and forestry survey, resource and environment remote sensing, disaster early warning and the like,and has great potential social value, economic value and benefits.

The invention discloses a license plate detection method based on an improved BING algorithm, and belongs to the field of license plate recognition. According to the invention, a BING detection windowis improved according to the specific shape of a license plate, and the BING detection features are enabled to be abundant, thereby enabling a recommended frame given by BING to completely comprise the whole license plate. Because the detection shape of BING in the invention changes, the method does not need to carry out the detection of images at different scales like an original method of BING,and can obtain a good detection result through employing a conventional pyramid algorithm.

The invention provides a texture mapping method in three-dimensional virtual human head and face modeling, and the method comprises the steps: cutting a front view and a side view of a real human face of a to-be-mapped person according to the position of a feature point through building a personalized human face model, and calculating the offset and the offset angle of each flush region; performing shear transformation on the cut side view according to the offset and the offset angle to obtain a side view after shear transformation; and introducing a Laplacian pyramid algorithm to fuse the mirror image of the cut front view, the mirror image of the side view after the shear transformation and the mirror image of the side view after the shear transformation, and mapping the fused image into the personalized face model to obtain a texture mapping face model of the to-be-mapped figure. Texture cracks can be effectively reduced under the condition of ensuring low data acquisition cost, so that the mapped texture mapping face model is more vivid.

The invention relates to a quality analysis method and system based on machine vision in new energy battery manufacturing. The invention aims to provide a quality analysis method and system based on machine vision in new energy battery manufacturing so as to detect the coating condition of a pole piece on line in real time, thereby ensuring the product quality. According to the technical scheme, the method comprises the following steps: S1, acquiring front and back images of a product; S2, finding out initial approximate position areas of the coating areas on the front and back images of the product through an image pyramid algorithm; S3, performing Gaussian smoothing filtering once in the approximate position area to remove noise signals; s4, calculating edge images in the X direction andthe Y direction of the coating area by using a Prewitt operator; S5, calculating a gradient direction and a derivative of each pixel point by using the edge values in the X direction and the Y direction; S6, according to the gradient direction and the derivative, pixels with small edge response are removed, and a continuous skeleton line with the maximum edge response is left. The method is suitable for the field of product quality detection.

The invention discloses a watermark detection method and system based on SURF and pyramid algorithms. The watermark detection method comprises the following steps: S1, scaling an original image to thesame size as a second sample image; s2, transforming the second sample image into the same scale space as the original image through an SURF algorithm; s3, positioning an ROI region in the transformed second sample image through a pyramid algorithm; s4, determining a watermark position; and S5, detecting the watermark. The method has the advantages of high matching precision, high matching speedand the like.

The invention provides a high-rise structure dynamic displacement measurement method based on unmanned aerial vehicle aerial photography. The method comprises the steps of acquiring a dynamic displacement video of the top end of a high-rise structure shot by an unmanned aerial vehicle; capturing a plurality of to-be-measured points of the first frame of dynamic displacement image at the top end of the high-rise structure as measurement points; using a pyramid algorithm to carry out displacement moving point capture on the multiple frames of dynamic displacement images to obtain a plurality of optical streams; calculating optical flow vectors of the plurality of optical flows by using an optical flow method to obtain an optical flow field; performing threshold segmentation on the optical flow field to divide the multiple frames of dynamic displacement images into backgrounds and moving targets; extracting a pixel coordinate of a moving target in each frame of dynamic displacement image, and calculating a displacement coordinate of a measurement point by taking the measurement point of the first frame of dynamic displacement image as a zero point coordinate to obtain dynamic displacement initial data of the top end of the high-rise structure; and correcting initial data of dynamic displacement of the top end of the high-rise structure by using suspension displacement compensation of the unmanned aerial vehicle. According to the invention, the problem of a large measurement result error caused by the fact that the camera is far away from the moving part of the measured high-rise structure can be solved.

The invention discloses a fast intelligent tracking method and device based on classified learning. A target image is quickly tracked, a corresponding control angle is determined through calculation,the control angle is transmitted to a motor control module, the platform is driven to move to enable the lens center to be aligned with the moving target all the time, and therefore the image trackingof the target is achieved. A quaternion method is introduced to carry out attitude calculation and carry out dynamic error compensation, so that the real-time performance and the stability of systemattitude sense are provided. And a pyramid algorithm is improved to layer the image, so that an optimal matching point is quickly found, and the algorithm speed is greatly improved. The method and thedevice are suitable for the ground detection, target identification, tracking and the like of the small industrial unmanned aerial vehicles.

The invention discloses a shearing wave image denoising method based on a cutoff window, which mainly solves the problem of large time complexity of a traditional shearing wave based on a first basis function algorithm and the problems of inflexible frequency band division and poor denoising effect of the traditional shearing wave based on a traditional Laplace pyramid algorithm, and comprises the following steps of: carrying out discrete Fourier transformation on a noisy image and the construction on a rectangular cutoff window or a round cutoff window; obtaining an approximate image througha rectangular central cutoff window or a round central cutoff window, carrying out direction division on the frequency bands of the rectangular cutoff window or the round cutoff window acting on the frequency domain of the noisy image by the shearing wave to obtain a coefficient matrix; afterwards carrying out hard threshold denoising on the coefficient matrix; and then reconstructing the denoised coefficient matrix and the approximate image by combining the rectangular cutoff window or the round cutoff window with the shearing wave to obtain the denoised image. The cutoff window constructed by the invention can obtain more ideal effect on image denoising and can be used for image analysis.

The invention provides a multi-layer template adaptive matching target tracking method for an infrared image, and belongs to the field of image processing. The method comprises the steps: obtaining an infrared image, and carrying out the image enhancement preprocessing of the infrared image, and obtaining a first layer of an image feature pyramid; manually marking a target position on the infrared image, namely giving a target frame of a first frame; extracting images of different layers in a target frame on the infrared image by using a feature pyramid algorithm: carrying out target search on each layer by using an SSDA template matching algorithm, frame-selecting the maximum possible position of the target, and recording confidence and target moving position information at the same time; judging the confidence coefficient through an error function, and if scale change occurs, determining a new template; if morphological transformation occurs, updating the template in time; and if shielding occurs, enlarging the search area. According to the method provided by the invention, the effective features of the target can still be stored when the target is subjected to scale change, form change and shielding, and good tracking is kept.

The invention brings forward a rapid and automatic mosaic technology of aerial video in a search and tracking system. The technology provided by the invention comprises the following steps: two continuous frames of color images in an aerial video sequence are converted to grayscale images; three-valued processing is carried out on the grayscale images by an image three-valued method, edge feature information images are extracted by an edge detection sobel operator and feature points are extracted by an Harris algorithm; feature point matching is completed by an optical flow pyramid algorithm; optimal homography matrix of the color images is calculated by combining a DLT algorithm and a RANSAC algorithm; panoramic image mosaics is completed after image correction of the color images by correction geometric transition matrix; and dynamic regulation of image three-valued upper and lower thresholds and a threshold of the edge detection sobel operator is carried out during the process of panoramic image mosaics. According to the invention, image mosaic accuracy and robustness are raised.

The invention discloses a rotating machinery fault diagnosis method based on multi-waveletorder ratio bispectrum analysis, comprising: step 1, inputting original signals to a pyramid algorithm formulti-wavelet transformation; step 2, according to a signal analysis theory, seeking an element function combination maximizing decomposed signal energy; step 3, maximizing the total energy of all elementfunction signals via a connection algorithm, and then solving rotating speed signals; step 4, performing equal-angle re-sampling on the rotating speed signals, and performing bispectrum analysis of the re-sampled signals to obtain order ratio spectra of vibrating signals; and step 5, analyzing the order ratio spectra of the vibrating signals, and performing fault diagnosis of rotating machinery based on the vibrating signals. The method improves the accuracy of fault diagnosis, does not need to install such hardware as an angle coding disc, a tracking filter, a rotating speed meter and the like, and reduces the dependency of the system on the installation space and the complexity of the system.

This invention relates to a method of non-contact detection and identification of the type of different substances and mixtures as well as determining their characteristics as concentration, hardness, etc. The method comprises irradiation of the inspected object by a wave pulse or a series of such pulses; reception (1), amplification (2) and analog-to-digital conversion (3) of the signal, formation of output detailing wavelet coefficients and output approximating wavelet coefficients by means of fast discrete wavelet transformation of the digitized signal, by means of Mallat's pyramidal algorithm and orthogonal base functions, filtering (7) of the output approximating and detailing wavelet coefficients up to preselected ones, comparison of the filtered approximating and / or detailing wavelet coefficients, in the capacity of recognition attributes with preselected respective reference coefficients by the classifying device (8), and according to the comparison result the presence and type is determined and / or the studied characteristics of the inspected object is determined.

The invention discloses a power grid slice updating method and a terminal. The method comprises the following steps: obtaining a change record of power grid equipment, combining slice geographic blocks of the change record through employing a pyramid slicing algorithm, and obtaining index change data of the slice geographic blocks; determining a display layer according to the slice geographic block, and calculating associated slice data before and after the power grid equipment data is changed in real time according to the index change data and the display layer, so as to obtain incremental slice data; and according to a file storage model of a pyramid algorithm, generating an incremental slice file from incremental slice data, and updating power grid slices based on the incremental slice file, and dynamically completing generation and pushing of a vector slice file online, so as to realize online updating of slice map service. The problem that off-line map cutting and publishing need to be carried out regularly according to the newest full-amount power grid data in the traditional slice map service is solved, and real-time updating of the power grid slice file is realized.

The invention discloses a method for detecting diseases and insect pests of apple trees based on a DNN network and a spot detection algorithm, and relates to the technical field of detection of fruit tree diseases and insect pests. And input the data set, update the weight matrix W and the bias parameter b through the forward propagation algorithm and the backward propagation algorithm of the neural network; step 2, use the Gauss pyramid algorithm to scale the collected image and then perform image segmentation, and the image Separate the foreground and background of the image; step 3, perform histogram equalization on the image segmented by the image, and enhance the feature points in the image; step 4, use the LOG algorithm to extract the feature points in the image, and then use the open operation to process , remove the noise, step 5: Input the feature points processed in step 4 into the trained DNN neural network for judgment, and identify whether there are diseases and insect pests on the leaves.

A fast intelligent tracking method and device based on classification learning. Fast tracking is carried out through the target image, and the corresponding control angle is determined by calculation, and then transmitted to the motor control module, which drives the platform to move so that the center of the lens is always aligned with the moving target, thereby realizing image tracking of the target. The quaternion method is introduced for attitude calculation and error dynamic compensation to provide real-time and stability of system attitude. Improve the pyramid algorithm to layer the image, so as to quickly find the optimal matching point and greatly improve the algorithm speed. It is suitable for ground detection, target recognition and tracking of small industrial UAVs.