Quantization Parameter Decision of Initial and Scene Change Frame in Real-Time H.264/AVC

In the first-I frame or scene-change frame, unsuitable quantization parameter (QP) has lower quality and more fluctuation. In this paper, we proposed an adaptive spatial algorithm to obtain a suitable QP of these frames in order to improve quality. We exploit a Laplacian of Gaussian (LoG) operator to analyze the video content to adaptively decide the initial QP. Then, the QP of scene-change frame is adaptively re-calculated by LoG operator when a scene-change frame is detected. Experimental results revealed that the proposed adaptive algorithm improve PSNR by 0.42 dB.