46 results about "Weighted median filtering" patented technology

The invention relates to a self-adapting multi-stage weighted median filtering algorithm applied to digital images. According to the method, the noise point in the image is judged according to the characteristics of the noise value, in addition, the noise point is marked, then, a self-adapting multi-stage weighted median filtering method is adopted for carrying out filtering processing on the noise point, and in addition, in the filtering algorithm realization process, the detected noise point and the treated noise point do not take part in the calculation. The self-adapting multi-stage weighted median filtering algorithm has the advantages that the noise detection correctness is high, the similarity of the processed pixel point to the original image signal is high, and the image detail remaining capability is good.

The invention discloses a binocular stereo matching method based on edge information. Firstly, performing color conversion on a left image and a right image acquired by Gaussian sampling through a Gaussian color model, introducing gradient information, using a neighborhood median as a threshold to replace a central pixel of a window, and counting the number of different corresponding bits in a Census transform value by using a Hamming distance so as to obtain a matching cost amount; secondly, converting the acquired image into an undirected graph, performing gradient operation by using a weighting function, constructing a minimum spanning tree, and performing cost aggregation by combining a cross-scale cost aggregation method to obtain a parallax value; generating a disparity map accordingto the obtained disparity value by using a winner-king strategy; and finally, carrying out region division on the pixel points by utilizing a super-pixel segmentation algorithm, and carrying out optimization processing on the disparity map by combining weighted median filtering, so that disparity information with relatively high precision can be obtained, and particularly, a relatively accurate disparity map can be obtained in an occlusion region and an edge information discontinuous region.

The invention relates to a stereo matching method based on a self-adaptive weight and local entropy, belonging to the technical field of binocular stereo vision. The method includes redefining the calculation of the window weight according to the gray similarity and the spatial similarity under a self-adaptive weight theory framework; introducing a data smoothing item into the cost function, and determining the penalty coefficient of the data smoothing item by applying the image local entropy; and performing parallax refinement by applying left-right consistency check and weighted median filtering on the initial parallax map obtained by cost aggregation to obtain a high-precision parallax image. The accurate parallax map can be obtained, and the parallax accuracy is improved.

The invention relates to an H.264 compressed domain realtime video object segmentation method based on motion feature. The method comprises the following steps: after spatial domain normalization of a motion vector field is completed, weighting median filtering of the normalized motion vector field is carried out; then, based on the three motion features of the motion vector field including amplitude, divergence and circuitation, the motion vector field is divided into a plurality of objects according to motion feature similarity through adopting an improved statistical region growing method. The experimental result of an MPEG-4 testing sequence shows that: when a CIF format video sequence is processed in a computer with a 2.0GHz dual-core CPU and 900M memory, the average processing time of each frame is 15 ms which meets the requirements of most real-time application at 30fps; moreover, excellent segmentation quality is obtained. Because the method only makes use of motion vector field information, the method can also be used in the motion object segmentation of an optic flow field.

The invention relates to field of radar signal processing technology and discloses a method for blanket jamming radar detection based on full-frequency band spectrum analysis, comprising: obtaining accumulated values of spectrum respectively corresponding to certain numbers of pulses with a radar scanning cycle; based on the accumulated values, calculating the median value and average value of each accumulated value, obtaining the value after weighted median filtering corresponding to pulse and composing a weighted median filtering result; confirming jamming region collection composed by a certain number of pulses within a radar canning cycle, which includes a certain number of jamming regions; obtaining orientation codes which respectively corresponding to pulses included jamming area, and confirming the corresponding jamming orientation, jamming intensity, jamming type and jamming frequency point. The method can realize power analysis of frequency spectrum on radar working band, which helps engineering project and improves anti-jamming capability of radar, with high recognition rate.

A method of modeling seismic wave-field data in order to suppress near-surface and sub-surface related multiple reflection signals is provided. The reflection signals include main primary reflection signals, main random noise signals, main multiple reflection signals, residual primary reflection signals, residual random noise signals, and residual multiple reflection signals. Main random noise signals are separated from the reflection signals using a frequency-wavenumber domain method to provide data having suppressed main random noise. Main primary reflection signals are separated from the data having suppressed main random noise using frequency-wavenumber filtering and weighted median filtering to provide data having suppressed main random noise and main primary reflections. Multiple reflection signals are modeled using parabolic path summation on the data having suppressed main random noise and main primary reflections.

The present disclosure relates to image filtering techniques for enhancing image quality. In one embodiment, two filtering techniques are applied on an image. Firstly, an adaptive weighted median filtering operation is performed on an acquired low contrast image corrupted by impulsive noise. Subsequently, a guided image filtering on the image obtained from adaptive weighted median filtering operation to de-blur and enhance the contrast that ultimately assures to preserve the edges of the images. In addition, the image filtering for enhancing image quality is enhanced by several variations of data adaptive guided image filtering and adaptive window sizes for guided image filtering techniques.

The invention discloses a CT image denoising method based on self-adaptive median filtering, belongs to the technical field of medical image processing, and is particularly suitable for CT image denoising of new coronal pneumonia. The method comprises the steps of selecting a square filtering window with the size of n * n, comparing the adaptive maximum value and the adaptive minimum value of the gray value in the window with the gray value f (i, j) of the current pixel point, judging whether the current pixel point is a suspected noise point or not according to a first threshold T0, and if so, further accurately judging whether the current pixel point is a noise point or not according to a second threshold T1, if the current pixel point is not the suspected noise point or the noise point, traversing the next pixel point in the window; processing noise points through a center weighted median filtering method; and finally, outputting the CT image after median filtering and denoising. Image details are better protected while image denoising is kept; the problem that deviation exists in information transmission of a traditional entropy weight method is corrected by improving the entropy weight method, the optimal weight of center weighted filtering is determined through contribution values of all evaluation indexes to the denoising effect, and therefore the optimal denoising effect is achieved.

The invention relates to a stereo matching method based on segmentation cross trees, and discloses a stereo matching method based on segmentation cross trees. The method comprises the following six steps: step one, preprocessing left and right stereo images by use of a Gaussian filtering technology, step two, calculating initial matching cost, step three, constructing a segmentation cross tree for each pixel point to be matched of the left and right stereo images, step four, aggregating the initial matching cost in a segmentation cross tree support area by use of a one-dimensional integration image acceleration technology, step five, calculating parallax errors according to a "winner-take-all" method, and step six, according to the parallax error results of the left and right stereo images, correcting the parallax error results respectively by use of a left-right consistency detection technology and a weight median filtering technology. The method provided by the invention is applied to stereo matching based on the segmentation cross trees.

The invention discloses a fabric defect detection method which specifically comprises the following steps: firstly, zooming a fabric defect image to be detected to 256 * 256 pixels, and then converting the fabric defect image into a gray level image; carrying out improved weighted median filtering processing on the fabric defect image obtained in the step 1; finally, using a K-means clustering tosegment the fabric defects. The filtering speed can be remarkably increased while the background textures of the fabric image are inhibited, and the defects of the fabric with different texture backgrounds can be effectively and accurately detected.

The invention relates to a graph cut stereo matching method based on adaptive weight, and belongs to the neighborhood of binocular stereoscopic vision technology. The method comprises firstly redefining the data items of an energy function according to grayscale similarity and spatial similarity under the theoretical framework of the adaptive weight, and using the gradient information of an imageas the smooth item of the energy function; then solving a model by using a graph cut theory and an [alpha] extended algorithm; and finally, subjecting a disparity map to disparity refinement by usingleft and right consistency test and weighted median filtering to obtain a high-precision disparity map. Four international standard stereo image pairs provided by the Middlebury website are used for testing and experiments show that the method can obtain a more accurate disparity map.

The invention provides a median filtering method and system based on vertical earthquake attenuation laws. The method comprises the following steps: collecting VSP (Vertical Seismic Profiling) data; selecting single component data according to the VSP data; carrying out zero-phasing treatment on the single component data, so as to obtain a profile treated through zero-phasing treatment; obtaining a time curve according to the profile treated through zero-phasing treatment; determining spherical spreading compensation TAR factors according to the profile treated through zero-phasing treatment and the time curve; pulling the profile treated through zero-phasing treatment to be homogeneous along the time curve, and obtaining a homogeneously pulled profile; carrying out directional weight median filtering on the homogeneously pulled profile according to the TAR factors of the VSP data. The median filtering method and system, provided by the invention, solve the problem of no reliable wave field separation method in VSP data processing used for well control.

The invention relates to a processing method and a system for reconstructing a foot point cloud model. The method comprises the following steps: obtaining a point cloud three-dimensional model reconstructed by matching a plurality of two-dimensional images; firstly, the 3D model of point cloud being divided into pieces by using the method of spatial hierarchical partition, and a simplified representation being calculated for each piece to obtain a simplified point cloud model; using the chord height difference method, connecting the front and back points of the detection point, calculating thedistance between the intermediate data point and the chord, comparing the distance with the given tolerance value, if the distance is greater than the tolerance value, determining that it is an abnormal point, and then deleting it, so as to remove the noise point in the place where the density is high and the curvature changes greatly in the point cloud model; then the weighted median filter being used to eliminate burrs and smooth the model; by calculating the length of each side of the uniform hole boundary, according to the calculated size of each included angle, the filling point coordinates being calculated, and the illegal judgment points falling outside the hole area being removed; repeating the process until a new filling point cannot be calculated and the hole filling ends.

The invention discloses a sequential Synthetic Aperture Radar (SAR) space-time neighborhood gauss weighting median filtering speckle noise suppression rapid algorithm, and provides a median filtering algorithm which performs gauss weighting according to time and space distance based on the condition that observation values of different time sequential SAR images are highly correlated with the observation values of a time neighborhood and a space neighborhood at the same time, and the larger the time interval is and the larger the space distance is, the less the correlation is. A section where the median is is pre-estimated according to statistic distribution characteristics of the SAR images, a median searching scope is reduced, and a median solving speed is accelerated. According to the invention, natural object features on the images can be kept well, and the speckle noise is suppressed. Meanwhile the influence of temporary natural objects such as vehicles and boats in the sequential images can be removed to a larger extent, and the sequential image speckle noise suppression problem is solved.

The invention discloses an SAR image edge detection method based on the anisotropic morphology direction ratio, and mainly solves a problem that the edge detection is poor in robustness of discrete strong-scattering points in the prior art. carrying out the anticlockwise rotation of Gaussian gamma-shaped double windows, and carrying out the weighted median filtering of the Gaussian gamma-shaped double windows; calculating the multi-direction edge response through the median filtering, and carrying out the multi-direction edge response; calculating the edge intensity mapping according to the edge response; extracting a maximum value point from the edge intensity mapping as a candidate edge point; carrying out the double-threshold decision making of the candidate edge point, and obtaining an edge detection result of a final image. According to the invention, the method achieves the improvement of the robustness of the SAR image edge detection for discrete strong-scattering points, meets the real-time requirements of the SAR image edge detection in a large scene, and can be used for the extraction of the SAR image edge in the big scene comprising the discrete strong-scattering points.

The invention discloses an iterative depth map structure restoration method for SSIM structural similarity based on RGB-D. The method includes: firstly, detecting the edge of an input depth map; and expanding the edge, marking the expanded area as a potential structure distortion area; judging whether each pixel point in the potential structure distortion area is distorted or not; generating a structural distortion measurement index; and enabling distortion pixel points to construct a recovery weight by adopting a product of a color image Gaussian weight and a structural distortion measurementindex, carrying out guide recovery through weighted median filtering, then carrying out guide filtering on a distortion region, continuously carrying out iteration on a finished result image according to the steps until a set iteration termination condition is met, and outputting a depth image to finish calculation. According to the method, iterative detection and recovery are carried out on thestructure distortion area of the depth map, so that relatively accurate structure information is obtained, denoising and edge preserving are carried out on the undistorted area of the structure, and finally, the depth map with a clear structure and a smooth depth value can be obtained.

According to a high-resolution SAR image segmentation method for improving the FCM through multi-stage cooperation, the theoretical basis of an FCM clustering algorithm and the limitation of the FCM clustering algorithm for high-resolution SAR image segmentation are analyzed, weighted median filtering improvement is conducted on the FCM in combination with the spatial neighborhood relation of pixels, aiming at the limitation that a current FCM segmentation result is prone to falling into local optimum, SA is improved to further optimize the FCM, the effectiveness of the improved and optimized algorithm is verified through experiments, finally, the segmentation result of the improved and optimized FCM clustering algorithm is used as the initial segmentation of maximum posterior probability superposition re-segmentation, the maximum posterior probability superposition SAR re-segmentation is adopted to segment the image, qualitative and quantitative comparative analysis is carried out on the segmentation results of the above methods to obtain a series of improvements, the SAR segmentation quality is obviously improved, details and contours of image edge areas are clear, segmentation is accurate, robustness and reliability are good, meanwhile, the resistance of the algorithm is enhanced, and the SAR segmentation quality and efficiency are greatly improved.

The invention discloses a wind generating set blade monitoring flaw detection method. Specific to the characteristics of fan blade acquisition images of an unmanned aerial vehicle, an unmanned aerial vehicle camera calibration method, a fast adaptive weighted median filtering algorithm, an image enhancement algorithm, blade surface fault feature extraction and selection, and blade surface fault classification monitoring and identification are used, and embedded into a man-machine interaction platform. Through online and offline tests, the platform can realize automatic identification and monitoring of sand holes, cracks, peeling and other defects, and the accuracy can reach 90% or above; compared with traditional monitoring means and other monitoring algorithms, the platform has higher accuracy and higher algorithm stability; and a new way is provided for nondestructive monitoring of wind driven generator blades.

The invention discloses a dual camera based automatic focusing method. The method includes the following steps: adopting a binocular camera that only has horizontal displacement to obtain a pair of binocular images at a default alignment plane, adopting a semi-global stereo matching method to solve the parallax of the binocular images, and partitioning the parallax to obtain a preliminary parallaxdiagram; correcting the parallax by using a weighted median filtering method, and removing a singular value in the parallax diagram to obtain a final parallax diagram; after a user selects a focusingarea, extracting a parallax value of the focusing area by using a focusing strategy in which a large screen priority mode and a close-range priority mode are combined, and calculating a focusing distance by using a trigonometric ranging method based on the parallax value and internal and external parameters of the binocular camera to achieve one-step focusing. The invention proposes the focusingstrategy in which the large screen priority mode and the close-range priority mode are combined for a single picture, the main body of the picture can be highlighted, and the one-step focusing can beachieved by using only a single pair of binocular images; and the method disclosed by the invention is small in calculation amount and good in real-time performance, and greatly improves the focusingspeed.

The invention discloses a Bayer image processing method and system, electronic equipment and a storage medium. The processing method comprises the following steps: acquiring a plurality of neighborhood pixel points of each target pixel point in a Bayer image; performing interpolation on the channels of the target pixel points corresponding to the other two color parts to obtain a first interpolation result set and a second interpolation result set; obtaining the edge direction of the target pixel point; obtaining a first weight of each first interpolation result in the first interpolation result set and a second weight of each second interpolation result in the second interpolation result set; and filtering the interpolation result according to the first weight and the second weight by adopting a weighted median filtering algorithm to obtain full-color pixel data of the target pixel point, and further generating a full-color image corresponding to the Bayer image. According to the method, different textures of the Bayer image can be effectively recovered, and pseudo-color and zipper effects generated by an existing interpolation technology are effectively reduced.

The invention discloses a mechanical arm intelligent control system based on multi-view stereoscopic vision. The system comprises a mechanical hand and a mechanical hand controller, wherein the mechanical hand controller is in signal connection with the mechanical hand. The periphery of the mechanical hand is fixedly provided with a binocular camera image acquisition unit acting on a target object, and the binocular camera image acquisition unit is connected with an upper computer through a data line; the binocular camera image acquisition unit adopts the improved algorithm combining cross-scale guided filtering with weighted median filtering parallax detailing, three-dimensional reconstruction of the target object is realized, coarse positioning of the target object is completed, and mechanical arm motion trajectory planning is carried out by adopting a 5-3-5 polynomial interpolation trajectory planning method; a monocular camera image acquisition unit is arranged at a clamping jaw ofthe mechanical hand and moves along with a mechanical arm to acquire an image; and the monocular camera image acquisition unit is in communication with the upper computer through a wireless transmission module, and the monocular camera image acquisition unit adopts a template matching method based on SURF for accurately positioning the target object.

The present disclosure relates to image filtering techniques for enhancing image quality. In one embodiment, two filtering techniques are applied on an image. Firstly, an adaptive weighted median filtering operation is performed on an acquired low contrast image corrupted by impulsive noise. Subsequently, a guided image filtering on the image obtained from adaptive weighted median filtering operation to de-blur and enhance the contrast that ultimately assures to preserve the edges of the images. In addition, the image filtering for enhancing image quality is enhanced by several variations of data adaptive guided image filtering and adaptive window sizes for guided image filtering techniques.

The invention discloses an improved feature stereo matching method based on binocular vision. The method comprises the following steps: carrying out the feature extraction of a left image and a right image after preprocessing, carrying out the feature matching, and obtaining an accurate matching point pair through screening; the obtained sparse matching point pairs are used as seed points, a one-dimensional search space is established according to parallax continuity and epipolar constraint criteria, a fast zero mean normalized cross correlation coefficient simplified by an integrogram is used as a similarity measurement function, region growth is realized through a bidirectional matching strategy, and the algorithm complexity is reduced while the matching accuracy is greatly improved; through sub-pixel fitting and weighted median filtering post-processing, the parallax precision is improved, and parallax step layering, noise and stripe phenomena are removed.

The invention relates to an infrared image noise removal method based on adaptive weighted median filtering. The method comprises the following steps: 1, establishing an infrared image self-adaptive median filtering noise removal method; and 2, distinguishing whether a processing object is noise or image details by utilizing an image gray variance, and finally realizing adaptive median filtering denoising of the infrared image through a center weighting mode, so that the method greatly ensures edge detail information of the image under the condition of removing background noise of the infraredimage, and has a relatively high application value.

The invention discloses an iterative depth map structure restoration method based on RGB-D SSIM structure similarity. First, the edge of the input depth map is detected, and the edge is expanded, and the expanded area is marked as a potential structural distortion area, and then the Each pixel in the potential structural distortion area judges whether it is distorted, and generates a structural distortion metric index. The distorted pixel uses the product of the Gaussian weight of the color image and the structural distortion metric index to construct a restoration weight, and performs guided restoration through a weighted median filter, and then The distorted area is guided and filtered, and the completed result map continues to iterate according to the above steps until the set termination iteration condition is met, and the depth map is output to end the calculation. The present invention obtains more accurate structural information by iteratively detecting and recovering the structurally distorted area of ​​the depth map, and at the same time denoises and preserves the edges of the structurally undistorted area, finally obtaining a depth map with a clear structure and smooth depth values .

The invention discloses a block matching and optical flow method combined foundation cloud picture motion prediction method, which comprises the steps of extracting a cloud region by utilizing cloud classification information for an acquired foundation all-sky image, calculating a main motion vector of the cloud region by adopting a block matching method, and taking a result as one of constraint items of an optical flow method; forming an optical flow method energy function jointly by the main motion vector, a brightness constant item, a global smooth item and a brightness gradient item; adding weighted median filtering to each layer of pyramid in the optical flow method cloud motion field calculation process, thus reducing the influence of a motion vector abnormal value, and realizing rapid and accurate prediction of cloud motion of a foundation all-sky image. Aiming at the problems of low block matching prediction precision, long calculation time of an optical flow method, light sensitivity and the like, the method of the invention improves the speed and precision of cloud motion prediction of an all-sky image by combining block matching with the optical flow method and fully utilizing the advantages of high block matching speed and high precision of the optical flow method.

The invention provides an automatic label content identification method based on a character two-dimensional code image, and the method comprises the steps: sequentially carrying out the denoising, binarization, positioning, error correction and decoding of the character two-dimensional code image on a steel label, and carrying out the denoising of the character two-dimensional code image through adaptive weighted median filtering, binarization is carried out on the character two-dimensional code image by adopting a binarization algorithm based on background gray level, and positioning is carried out on the character two-dimensional code image by utilizing a positioning algorithm based on a convolutional neural network; and when the character two-dimensional code image is distorted, correcting the character two-dimensional code image by adopting a neighbor method. By using the method of the invention, the influence of the storage and transportation process on the character two-dimensional code image on the steel label can be solved, and the related information of the steel can be accurately obtained.

The invention provides a real-time weighted median filtering method based on bilateral grids. This method can be briefly summarized as the following five steps: the user is given an input image and a boundary image; a four-dimensional bilateral network with four coordinate axes (u, v, z, a) is constructed based on the input image and boundary image. grid; according to the constructed four-dimensional bilateral grid, perform Gaussian filtering operations along the three coordinate axes of u, v, and z to obtain a smooth weighted bilateral grid; according to the smoothed weighted bilateral grid, along the a coordinate axis Perform an accumulation operation to obtain the accumulated three-dimensional histogram. According to the accumulated three-dimensional histogram and based on the linear interpolation method, the final weighted median filtered output image is obtained. Benefiting from the advantages of the bilateral grid, the method of the present invention can maintain good edge information while smoothing the image. Each step of the method of the present invention can be processed in parallel in the graphics card hardware, effectively improving the running speed.