Image compression by variable block truncation coding with optimal threshold
Mohamed S. Kamel, C.T. Sun, L. Guan · IEEE Transactions on Signal Processing · 1991
A variable block truncation coding (BTC) algorithm is proposed for image compression. It is shown that there exists an optimal threshold for the quantization in BTC algorithms (fixed and variable) that minimizes the errors. Compared to the fixed BTC (fBTC), the variable BTC (vBTC) gives better performance on all the tested images. The use of vBTC with optimal threshold leads to a reduction of the error in the reconstructed images by almost 40% of the error in the reconstructed images obtained by fBTC. This enhanced performance suggests that the vBTC with optimal threshold is a better alternative to the fixed block truncation coding.>