Image real-time processing method

Abstract

The invention provides an image real-time processing method. The image real-time processing method comprises projecting multi-scale image characteristic space of an image and classifying a target and a background sample in a next frame image through a classifier; searching within a local range through a target vision model and searching an optimal target position through a regularization cross correlation coefficient. The image real-time processing method improves the tracking accuracy and retains the image processing instantaneity to some extent.

Description

Technical field [0001] The invention relates to image processing, in particular to a method for tracking a target in an image. Background technique [0002] Motion tracking is the core component of the intelligent transportation system. It is used to fully automatically locate vehicle information from the captured vehicle images, including license plate images, model recognition, and accident identification, making computerized vehicle monitoring and management a reality. In the traditional moving target tracking method, the final tracking target obtained by the classification process of the target and the background is only the target sample that is most similar to the target and the most dissimilar to the background at the same time, indicating that the determination by the classifier is not necessarily the same as that obtained only based on the target vision Is consistent. Although the robustness is achieved, the most accurate target is not tracked, and the positioning accur...

AI Technical Summary

Problems solved by technology

Although robustness has been achieved, the most accurate target has not been tracked, and the positioning accuracy needs to be further improved

Therefore, for the above-mentioned problems existing in related technologies, effective solutions have not yet been proposed

Images