Display region segmentation from a computer screen image using closed-contour tracking

摘要

Based on closed-contour tracking, a real-time approach for the display region segmentation from a computer screen image is presented in this paper. This approach provides a significant preprocessing step while computer vision is perceiving digital-world information through a computer screen. To segment the display region, a closed contour between the screen frame and display region should be found first in our approach, in which a grayscale image containing the whole screen area is required. Because of some inherent characteristics embedded in a computer screen image, a mean-C adaptive thresholding algorithm is used to obtain a binarized image that contains the wanted closed-contour boundary. Then a parallel thinning algorithm is applied to reduce the thickness of the closed contour to be 1-pixel wide. Finally, our closed-contour tracking algorithm is used to perform the display region segmentation. Experiments confirm the feasibility of the proposed approach.