36 results about "Texture Descriptor" patented technology

Provided are methods for managing texture data. The methods include preloading a first plurality of texture descriptor values from a memory location in a first buffer located in a first logic block, wherein the first buffer is further configured to receive data corresponding to non-texture functions performed in the first logic block and preloading the first plurality of texture descriptor values from a memory location into a second buffer in a second logic block if the first buffer is full. The methods further include utilizing the first plurality of texture descriptor values, within the second logic block, to perform a shader calculation, and loading, dynamically, a second plurality of texture descriptor values from memory into the first buffer, wherein the first logic block requires additional data. Additionally, the methods can include writing, if the first buffer is full, the second plurality of texture descriptor values over a portion of the first plurality of texture descriptor values.

The invention discloses a cable defect detection optimization method based on a gray scale run matrix, and relates to the field of artificial intelligence. Comprising the steps of obtaining multiple regional images; calculating a roughness coefficient of each pixel point in each regional image according to each pixel point in each regional image and gray levels of pixel points in eight neighborhoods of each pixel point, further obtaining a roughness coefficient level image, and obtaining a gray level run length matrix corresponding to each regional image; calculating the roughness degree of each region image in each direction; the pixel points not smaller than the run threshold are marked in the direction with the maximum roughness degree, all marked areas are obtained, and the marked areas are combined to obtain a rough area; and judging each obtained rough region to obtain all cable defect regions. The texture features of the weak unsmooth area are extracted based on the improved gray scale run matrix, a better extraction effect can be obtained compared with a conventional texture descriptor, and the defect detection precision is effectively improved.

The invention provides an image processing method for detecting a pulmonary tuberculosis focus in a chest X-ray DR film. The image processing method comprises the following steps of extracting a lung area image through an active shape model; extracting SURF key feature points of a lung area image; extracting texture descriptors of the key feature points; assigning the kind of the key feature points; performing clustering and combination on the texture descriptors according to a k-means clustering algorithm, and performing classified making on a combination result; setting a group of dimension values, and combining the texture marks of the lung area image according to a sequence from a least dimension value to a minimum dimension value; and deleting the class of which the number of samples is smaller than 5% of the total number in the combining result, thereby only reserving main classes in the combination result, eliminating abnormities and obtaining an image processing result. The image processing method can be used for automatically processing the DR films. The image processing method is applied for screening the pulmonary tuberculosis. The lung DR films which may represent the pulmonary tuberculosis can be screened. The image processing method can be used for large-number pulmonary tuberculosis screening. The image processing method has advantages of high screening efficiency and objective and stable conclusion.

The invention is applicable to the field of medical image processing, and provides a dividing method based on the CT (Computed Tomography) image liver tumor focus. The method comprises the following steps of preprocessing an initial CT image; completing ROI selection on the suspected focus in an interactive way by aiming at the preprocessed CT image; performing texture description on the ROI based on a CT value, and obtaining probability spectrums of the ROI through weighted computation of texture descriptors; building a big data priori knowledge base, determining a focus region probability spectrum threshold value, and dividing the suspected focus based on the threshold value; and completing the volume statistics and the quantization output on the focus. The method has the advantages that a plurality of texture descriptor probability spectrums are obtained based on stable and comparable CT value calculation; meanwhile, the threshold value of the texture descriptor probability spectrums is obtained through manual division calculation based on the priori knowledge base; and the accuracy and the timeliness of the multi-domain liver tumor focus division in the CT image can be realized.

Techniques for detecting objects across images captured by camera devices. Embodiments capture, using first and second camera devices, first and second pluralities of images, respectively. First and second reference images are captured using the first and second camera devices. Color descriptors are extracted from the first plurality of images and the second plurality of images, and texture descriptors are extracted from the first plurality of images and the second plurality of images. Embodiments model a first color subspace and a second color subspace for the first camera device and the second camera device, respectively, based on the first and second pluralities of images and the first and second reference images. A data model for identifying objects appearing in images captured using the first and second camera devices is generated, based on the extracted color descriptors, texture descriptors and the first and second color subspaces.

The invention discloses a method for detecting image characters in natural scene and relates to a method for adopting a texture descriptor LHBP to describe the texture character of an image and adopting a multi-dimension tropistic wave filtering method to detect characters in the image, thereby solving the problem that the character detection method based on texture has complex requirement on illumination; and the change of contrast gradient between the character and the background has great influence on detection. The obtained LHBP tropistic texture code and the corresponding code produced according to the change of position weight are obtained; a character region is determined through a multi-dimension tropistic analytic method. The method adopts a mode of extracting local texture on multi-dimension wavelet character by the LHBP texture descriptor, can filter out the influence of complexity and the change situation of the contrast ratio between the character and the background, effectively extracts the texture character of the character region, utilizes the texture direction property of the character region to determine the final character region and has good robustness in complex illumination, the change of the contrast gradient between the character and the background, the change of the size and the stroke thickness of the character and the like.

The invention belongs to the technical field of image processing, and relates to a colored textile fabric color difference degree detection method which is widely applied to the fields of artificial intelligence, color analysis, pattern recognition, intelligent detection and the like. According to the method, color space conversion is firstly carried out on an image, then global color features andlocal texture descriptors are extracted from HSV and Lab color spaces respectively, and finally corresponding features are fused to obtain mixed color space color representation features. According to the method, the defect that colors are described in a single color space is overcome, the color depicting capacity of the mixed color space is improved, the color difference degree is accurately detected, and the method has universality for different colored spun fabrics.

A method of describing texture of a group of frames (GoF) or a group of pictures (GoP) using homogeneous texture descriptors, and a method and apparatus for retrieving a GoF/GoP using the texture description method are provided. The texture description method includes: generating homogeneous texture descriptors of all frames constituting the GoF or all pictures constituting the GoF; and expressing the GoF or GoP using a predetermined representative homogeneous texture descriptor corresponding to one frame or picture to reduce the amount of data. The GoF/GoP retrieval method includes: establishing a database of homogeneous texture descriptors of a plurality of GoFs's or GoPs's, each GoF or GoP being expressed by a predetermined representative homogeneous texture descriptor corresponding to one frame or picture to reduce the amount of data; generating a homogeneous texture descriptor corresponding to one frame or picture of a query GoF or GoP when the query GoF or GoP is input; searching homogeneous texture descriptors that are similar to the homogeneous texture descriptor of the query GoF or GoP in the database; and retrieving GoFs's or GoPs's corresponding to the searched similar homogeneous texture descriptors and arranging GoFs's or GoPs's in the order of degree of similarity. Therefore, the texture of images can be more accurately expressed, and an image can be more efficiently and rapidly retrieved.

The invention provides a local texture description method based on a local grouping comparison mode column diagram. The method specifically comprises the following steps: step 1, selecting q supporting regions on the basis of an interest region; step 2, normalizing the supporting regions into circular regions; step 3, dividing circular images into P parts through convergence strategies based on mean value ordering; step 4, calculating a local grouping comparison mode of pixel coordinate points inside the circular images; step 5, counting a local grouping comparison mode column diagram in a local characteristic region according to the suffix of the local grouping comparison mode, so as to form local texture description of a single supporting region; step 6, cascading the local texture descriptions of the supporting regions to obtain a local grouping comparison mode column diagram for characteristic regions, and forming a local texture descriptor. According to the method, actually, a gray change column diagram for pixel point neighbourhood pixels with consistent gray values is calculated, and the descriptor formed by the method is strong in discrimination performance and has excellent robustness in illumination transformation and geometric transformation.

The invention discloses a decomposition method of image texture and structure based on a total variational model of a guide map, belonging to the technical field of image processing. According to theinvention, firstly, the local main structure of the image is reconstructed based on the guide image filter, then the texture descriptor is calculated according to the reconstructed local structure image, finally, the accuracy and computational efficiency of texture and structure decomposition are improved by combining the multi-scale total variational model and block translation method. The technical scheme of the invention can obtain better texture/structure decomposition effect for noisy images, the textures and structures of different scales can also be decomposed accurately, and the decomposed structure layers can keep the original shading of the image, avoiding the blurring of the structure caused by local smoothing or the color block effect caused by the global optimization method. In addition, the technical proposal of the invention needs to extract simple features, and does not depend on learning a large number of image samples.

An image texture classification method based on jump subdivision local pattern includes:firstly, skipping local difference counting feature (JLDCP) information is extracted from the second-order difference counting feature and diagonal difference counting feature, then the subdivision complete local binary feature (RCLBP) is extracted from the symbol information and size information of the subdivision complete local binary feature, finally, the texture descriptors (JRLP) of skipping and subdivision local patterns are obtained by connecting the skipping local difference counting feature (JLDCP)and the subdivision complete local binary feature (RCLBP). The invention has the beneficial effects that the invention has robustness to image noise, rotation, scale and illumination change and the like.

The invention provides an extraction method of a histogram texture descriptor in a muti-contrast mode, comprising the following steps: carrying out image quantization; separating the quantized image into a positive matrix, a negative matrix and an equivalent matrix; and carrying out histogram calculation according to the three matrixes to obtain the histogram descriptor in a local mode of the image. The technical proposal provided by the invention can achieve efficient and simple extraction of textural features and well express the texture of an SAR image.

According to the fast coding unit division method for H.266/VVC intra-frame prediction coding, the texture complexity of a current coding CU is judged according to a texture descriptor, if the current coding CU is a flat region, division of the current CU is terminated, quadtree division is carried out on a CU block with the size of 64 * 64 to obtain a CU block with the size of 32 * 32, horizontal and vertical activity degrees are calculated, and the CU block with the size of 32 * 32 is obtained. If the difference between the horizontal activity degree and the vertical activity degree is smaller than a set threshold value, a VTM default mode is adopted, if the horizontal activity degree is larger than the vertical activity degree, texture activity degrees of quadtree division, horizontal binary tree division and horizontal ternary tree division are calculated, and if the vertical activity degree is larger than the horizontal activity degree, texture activity degrees of quadtree division, horizontal binary tree division and horizontal ternary tree division are calculated; the texture activity degrees of quadtree division, vertical binary tree division and vertical ternary tree division are calculated, the texture activity degrees are judged, a corresponding division mode is selected, and otherwise, a VTM default mode is adopted; according to the method, flat CU block division can be terminated in advance, a large amount of cost calculation is skipped, and the coding time is simply and effectively saved.

A method for retrieving an image texture descriptor for describing texture features of an image, including the steps of (a) filtering input images using predetermined filters having different orientation coefficients, (b) projecting the filtered images onto axes of each predetermined direction to obtain data groups consisting of averages of each directional pixel values, (c) selecting candidate data groups among the data groups by a predetermined classification method, (d) determining a plurality of indicators based on orientation coefficients of the filters used in filtering the candidate data groups, and (e) determining the plurality of indicators as the texture descriptor of the image. The texture descriptors which allow kinds of texture structure present in an image to be perceptually captured can be retrieved.

A digital image texture analyzing method uses the mean and variance of the pixel values of an original image. Further, filtered images are obtained by filtering the original image using predetermined filters, each having a unique combination of one of m scales and one of n orientations, where m and n are predetermined positive integers, and the means and variances of the respective filtered images are calculate. A texture descriptor is obtained having the mean and variance of the pixel values of the original image and the means and variances of the respective filtered images obtained in the step, as texture features. The texture analyzing method allows image textures to be more accurately analyzed and compared for similarity, even when an image is rotated and/or enlarged or reduced relative to another image.