Multiple-scale object tracking method using adaptive characteristic fusion

Abstract

The invention discloses a multiple-scale object tracking method using adaptive characteristic fusion; the method comprises the following steps: a, a feature extraction step: reading image and initialization object positions, extracting HOG features and CN features of an object image, calculating color information entropy of the image, and carrying out adaptive characteristic fusion; b, a multi-scale classifier training step: using a cosine window function to filter a characteristic matrix, multi-scale zooming the characteristic matrix, converting the multi-scale characteristic matrix into Fourier expansion for calculation, thus obtaining classifier models of various scales; c, an object detection step: reading the next frame of video image, extracting features, converting the features intothe Fourier expansion domain, using a multi-scale model to calculate the optimal object position, building a Bayes scale estimation framework, and solving the object optimal scale; d, a model updating step: re-training classifiers for newly detected object positions, and updating models of original classifiers and newly obtained classifiers according to certain linear proportion. The method can effectively improve the feature expression ability, so the object scale estimation can be more accurate, thus greatly improving the tracing precision.

Description

technical field [0001] The invention belongs to the field of computer vision, is an important application technology in the field of target tracking, and in particular relates to a multi-scale target tracking method based on adaptive feature fusion. Background technique [0002] With the rapid development of computer technology and electronic information technology, people use computers to simulate human visual system for information collection, analysis and processing, so that computers can perceive the external world and perform related processing. It allows the computer to observe, recognize, and understand images by performing relevant processing on images, helping people better process massive amounts of data information, liberating humans from tedious mechanical work, and accelerating the construction of social informatization process. [0003] Object tracking is a very important research direction in the field of computer vision, which includes many cutting-edge tech...

AI Technical Summary

Problems solved by technology

[0006] The problem to be solved by the technology of the present invention is: firstly, in view of the problem that the traditional tracking algorithm cannot adaptively change the feature fusion weight according to the different characteristics of the image during feature fusion, an adaptive feature fusion method is proposed

Images