MS-LBP (Multi-scale Local Binary Pattern) feature extraction method and system based on path integral

Abstract

The invention relates to a MS-LBP (Multi-scale Local Binary Pattern) feature extraction method and a system based on path integral. The method comprises steps: a cross-scale path set and multiple high pass filters are built, and a path and a high pass filter with the same length are combined to form a feature parameter group; as for each pixel point, according to each feature parameter group, a multi-scale path integral local binary pattern (pi-LBP) feature value is extracted; as for pi-LBP feature values of all pixel points in each feature parameter group, a histogram is built, and all histograms are serially connected to form a final feature value of the image. Through building different cross-scale paths and high pass filters, the texture correlation among different scales is extracted, the problem of single change of LBP coding can be solved, more effective features can be extracted through building the different paths and filters, and the feature recognition ability is thus improved.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a feature extraction method and system based on path integral MS-LBP Background technique [0002] Digital image is an important medium for information transmission and one of the main ways for human beings to obtain information. With the rapid development of the Internet, a large amount of image information will be collected every moment, so computers are needed to help humans obtain the information carried by these images. In order to effectively express the essential information of the image, it is necessary to perform feature extraction on the original image. The quality of feature extraction will directly affect the design and performance of image analysis. Therefore, feature extraction is a key issue in the field of computer vision. [0003] The currently extracted image features are mainly divided into global features and local features. The global feat...

AI Technical Summary

Problems solved by technology

[0006] In view of the above problems, the present invention provides a MS-LBP feature extraction method and system based on path integrals. By constructing different cross-scale paths and high-pass filters, the texture correlation between different scales is extracted, and the single change of LBP coding is solved. For the problem, more effective features can be extracted by constructing different paths and filters, thus improving the recognition ability of features

Images