Automatic Camera Calibration Technique and its Application in Virtual Advertisement Insertion System

This paper has proposed a novel approach to automatic camera calibration for the virtual advertisement insertion system. The presented technique can extract feature lines in the field automatically, through which the internal and exterior parameters of camera can be obtained. Firstly, the adaptive segmentation of color image is carried out to the football field images. Secondly, the feature lines are extracted by applying a multiple Hough transforms to the segmented images. The feature points in the image space which are corresponding to the reference coordinate of world are generated automatically from the intersections of the virtual grid-lines, which are used for automatic camera calibration. Then the advertisements can be inserted into the soccer field in the appointed positions according to these parameters. Experimental results have shown that this method can meet the needs of the virtual advertisement insertion system and be put into use in real application scenarios.