Compression ratio boundaries for predictive signal compression
Oleg S. Pianykh, John M. Tyler · IEEE Transactions on Image Processing · 2001
Predictive regressional models like DPCM are widely used in digital signal compression. This paper analyzes the relationship that exists between the predictive model fitness and the resultant reduction of the first-order signal entropy, and finds the lossless compression ratio C as a function of predictive model correlation rho.