- Volume 26 Issue 11A
Image and video compression requires quantization error model of DCT coefficients for post processing, restoration or transcoding. Once DCT coefficients are quantized, it is impossible to recover the original distribution. We assume that the original probability density function (pdf) is the Laplacian function. We calculate the variance of the quantized variable, and estimate the variance of the DCT coefficients. We can confirm that the proposed method enhances the accuracy of the quantization error estimation.