Implement the transform and quantization portions of the JPEG standard. For coding the labels use an arithmetic coder instead of the modified Huffman code described in this chapter.
(a) Encode the Sena image using this transform coder at rates of (approximately) 0.25, 0.5, and 0.75 bits per pixel. Compute the mean squared error at each rate and plot the rate versus the mse.
(b) Repeat part (a) using one of the public domain implementations of JPEG.
(c) Compare the plots obtained using the two coders and comment on the relative performance of the coders.