While watching a video human visual system gives more attention on the foreground objects than background objects. That is, human vision system pays more attention to the region of interest, such as the human faces in the video content. Most of the video steganography algorithms embed secret information in video by considering every part of the video frames with equal importance. So the capacity of the steganography could not be increased to maintain low visual distortion. The proposed system will detect the foreground and it will allocate different bit rates for different regions. By doing this the visual distortion will be maintained and the capacity can be increased up to 33%.