Hybrid statistical modeling-based texture image retrieval method

Abstract

The invention discloses an amplitude phase hybrid modeling-based image retrieval method. The method comprises the following steps of: firstly, calculating an amplitude and a relative phase of a transformed high-frequency sub-band by utilizing a DD-DTCWT transformation and decomposition image with a favorable directivity characteristic; secondly, carrying out statistical model modeling on the amplitude by applying a Cauchy distribution function so as to obtain a position parameter gamma and a scale parameter phi; carrying out statistical model modeling on the relative phase by applying a Vonn distribution function so as to obtain a position parameter mu and a scale parameter lambda; and finally, taking gamma, phi, mu and lambda as features of each image to be applied to image retrieval. Experiment results prove that gamma, phi, mu and lambda are taken as features of each image, so that dimensionalities of the features are effectively reduced, the time distributed by similarity calculation is shortened, and the average retrieval rate is improved.

Description

technical field [0001] The invention belongs to the technical field of digital image retrieval, relates to a content-based image retrieval method, in particular to a texture image retrieval method based on mixed statistical modeling. Background technique [0002] Today's society is a rapidly developing society, with increasingly developed technology and high-speed information circulation. At the same time, with the continuous development of Internet technology, the era of big data has come, and huge amounts of information are pouring in all the time, including a large amount of image information. People can capture images anytime and anywhere to form image information, and spread them to the Internet instantly. Digital images are growing at an alarming rate. The current problem is how to efficiently and accurately retrieve the huge image database. [0003] The early traditional text-based image retrieval method (TBIR) is a retrieval process based on word matching, which sea...

AI Technical Summary

Problems solved by technology

With the gradual expansion of the amount of image information, TBIR technology encounters severe problems: text annotation is difficult to describe more and more abundant image content in an all-round and accurate manner, and content-based image retrieval technology (CBIR) emerges as the times require.

The currently proposed CBIR technology algorithms are relatively cumbersome, difficult to implement, and time-complexity is high. Therefore, finding a CBIR technology with low complexity, fast operation speed, and good feature extraction effect is currently a topic that needs in-depth research.

Images