Evolved transforms improve image compression
Frank W. Moore · SPIE Newsroom · 2009
A novel technique for optimizing compression outperforms wavelets for data subject to quantization or thresholding. Many image processing applications, including JPEG20001 and fingerprint compression,2 use wavelets. By redistributing an im-age’s energy into a set of lower-resolution trend subimages, dis-crete wavelet transforms (DWTs) can significantly reduce the number of bits required for representation. Quantization (ap-proximating a signal with a smaller number of bits) and thresh-olding (setting all but the largest transform values to zero) may achieve additional compression. However, these techniques in-troduce irreversible information loss, since the mean squared error (MSE) of reconstructed images increases proportionately with both techniques. One approach that may circumvent this problem is evolu-tionary computation (EC), which uses fitness-based selection