Color compensation method based on illumination difference

Abstract

The present invention provides a color compensation method based on illumination difference. The method comprises the following steps: transforming a source image and a target image from a RGB color space to a Lab color space; after the source image and the target image are transformed to the Lab color space, calculating the color statistics characteristics of each channel of the two images; transplanting the color characteristics of one image into the other image, when a CCT model between each pair of adjacent cameras is calculated in a preprocessing phase, obtaining the CCT model according to the different observation results in the two camera images of the same object; and when there is no matching object, performing initialization of the CCT model through adoption of the color characteristics of the whole image in the camera visual fields. In the condition of ensuring that target characteristics are not lost, the color compensation method based on illumination difference realizes color transformation of the common object color transformation in the visual fields, and on the basis of a reliable characteristic extraction algorithm, the color compensation method based on illumination difference can effectively perform color compensation, add the reliability of appearance association and increase little time load.

Description

technical field [0001] The invention belongs to the technical field of image processing and relates to a color compensation method based on illumination difference. Background technique [0002] Visual object tracking is a hot issue in the field of machine vision research. Its main task is to obtain the trajectory of the object of interest in the video sequence. It has been widely used in video surveillance, intelligent transportation and other research fields. At present, challenging problems still need to be solved urgently: 1) internal factors such as pose changes, scale changes, and deformations of the tracking target itself; 2) external factors such as brightness changes, partial occlusions, and complex backgrounds in the tracking process. [0003] For a multi-camera surveillance network, a cross-camera multi-target tracking solution with overlapping fields of view can identify targets as they move from one camera perspective to another. [0004] However, the previous ...

AI Technical Summary

Problems solved by technology

[0004] However, the previous methods often established an appearance model based on color information to identify targets, because the color information of pedestrians’ clothing can usually represent the unique identity of the object to be matched under specific conditions with spatio-temporal information constraints, and the collected color information often Due to overexposure, backlight and highlights, etc., the effect will be too dark or too bright, which will greatly reduce the tracking and matching accuracy of the target.

Images