An Illumination Robust Color Space Construction Method for Visual Tracking

Abstract

The invention discloses a method for establishing a visual tracking color space with illumination robustness. The method comprises the steps of providing a method for performing interframe maintenance on a component H according to quantitative transformation formulae of a red-green-blue (RGB) space and a hue-saturation-intensity (HSI) space; classifying image pixel points into two major categories according to the presence or absence of color information, respectively correcting and restraining light sensitive components of the two major categories of image pixel points in the HSI space, and giving a specific operation method implemented in the RGB space so as to establish the new color space having robustness for illumination; applying the established new color space into a classic visual tracking algorithm. The method is capable of being used for enabling the traditional tracking algorithm to maintain higher stability and accuracy; only linear change is performed on the color space without involving the modification for the tracking algorithm, so that the whole calculation complexity of the method is greatly reduced; furthermore, the method has the characteristics of being low in calculation complexity and good in real-time performance.

Description

technical field [0001] The invention belongs to the field of optical signal processing and digital image processing, and in particular relates to a color space construction method based on the principle of hue preservation and robust to illumination changes. Background technique [0002] The research on tracking algorithms that are robust to illumination changes has long occupied a place in optical signal processing applications (such as military target strikes, video surveillance, and robot vision, etc.). For visual tracking problems based on target feature distribution, such as vehicles ([1] Wang Jinhua, Cao Jie. Research on moving vehicle tracking algorithm based on feature points [J]. Electric Automation. 2011 (06)), face tracking ([2] Xiao Bing, Wang Yinghui. A Review of Face Recognition Research [J]. Computer Application Research. 2005 (08)). Changes in light intensity caused by tree shade, weather, and camera parameters often cause drift in tracking results. Most of ...

AI Technical Summary

Problems solved by technology

The former uses the color state of the current frame of the target to predict and update the color model, so as to ensure that the next frame has a certain degree of adaptability to changes in color information, but this type of method is limited by the update speed and cannot cope with rapid changes in color information; the latter It is to eliminate the influence of the change probability density of the color distribution by calculating the constant optical flow field, which has certain adaptability to the light change speed, but its calculation amount is still large, and it cannot be processed in real time.

Images