Real-Time Video Enhancement for Moving-Object in the Dusk Environment Based on Pixel Classification
Chao-Ho Chen, Che-Ping Su, Tsong-Yi Chen, Da-Jinn Wang · 2010
In this paper, we proposed a novel video enhancement for moving-objects in the dusk environment based on mixed-mode processing which involves de-noising, moving object detection, and image enhancement. The basic strategy is to first calculate each pixel's motion vector and then check the magnitude of this motion vector to determine that the pixel belongs to foreground (i.e., moving object) or background. Then, the pixel of foreground is processed with the temporal average filter to further check if it belongs to the moving object by motion detection. Finally, the extracted moving object is enhanced by using Bilateral filter and Retinex algorithm. Experimental results show that the proposed method is able to divide the frame into foreground and background and the foreground has a significantly clear and high contrast appearance in the low-light environment.