Distributions of the DCT coefficient for Watermark Detection

In this paper, we derive the distribution of the DCT coefficients for a fixed block variance, and invoke the central limit theorem to show that this is approximately Gaussian. Then, by allowing the block variance to have a statistical distribution itself, we analyze when the DCT coefficients would follow a Laplacian distribution. Such knowledge would be useful, for instance, in quantizer design, information hiding capacity and Watermark Detection.