Weighted gradient direction co-occurrence matrix textural feature extraction method

Abstract

The invention discloses a weighted gradient direction co-occurrence matrix textural feature extraction method. The method comprises the steps that a set of displacement factors are pre-defined; a gradient amplitude image and a gradient direction image are calculated for to-be-analyzed images; the gradient amplitude image is coded by using a local binary pattern algorithm to obtain the gradient amplitude coding value of each pixel; for each pre-defined displacement factor, on the basis of the gradient direction image, a co-occurrence matrix is calculated with the gradient amplitude coding values as the weights, and weighted gradient direction co-occurrence matrixes are obtained; vectorization and normalization processing are conducted on all the weighted gradient direction co-occurrence matrixes to obtain weighted gradient direction co-occurrence matrix textural features. The method overcomes the limitation that a traditional co-occurrence matrix textural feature extraction method onlycarries out statistics on single image information, and the purpose of improving the target description ability is achieved.

Description

technical field [0001] The invention relates to image processing technology, in particular to a method for extracting texture features of a weighted gradient direction co-occurrence matrix. Background technique [0002] Image features reveal the essential properties of images, and image feature extraction technology has always been an important research content in the field of image applications. Image feature extraction is essentially a process of eliminating redundant information. It is the premise of subsequent image segmentation, recognition, classification and other operations. It can effectively improve the accuracy of detection and recognition in subsequent applications, and can effectively reduce the amount of calculation and improve calculating speed. [0003] Texture is a ubiquitous feature on the surface of an object, which reflects the unique structure arrangement information of the object surface, and has strong robustness to brightness and color changes. Imag...

AI Technical Summary

Problems solved by technology

However, the existing gradient information-based co-occurrence matrix texture feature extraction methods usually only count single gradient magnitude information or gradient angle information, and do not consider the superiority of multi-information to image feature description.

Images