Bit-depth scalable video coding using error residual correction

We propose a new algorithm for improving the coding efficiency of bit-depth scalable video coding. While coding the enhancement layers in bit-depth scalable coding under scalable video coding framework, we apply a correction to the error residuals generated after the inverse tone mapping process. The correction factor is towards reducing the residual values close to zero and this leads to reduction in bitrate. While decoding, we compensate for the correction applied, by applying a filter based on laplacian kernel to the reconstructed high bit-depth images. The application of the laplacian filter helps in improving the quality (PSNR). We have implemented our algorithm in JSVM 8.12 reference software and performed experiments on different SVT and VIPER video sequences. Our results on an average, show a overall savings in bitrate (BDBR) of about 3.97% and a quality (BDPSNR) improvement of 0.19dB, when compared to existing JSVM 8.12 implementation.