Experiments in simple one-dimensional lossy image compression schemes

Abstract

The paper examines some one-dimensional lossy image compression schemes. To perform one-dimensional compression, an image is scanned into a one-dimensional array which is then compressed. Three types of scanning are considered: raster scan, Hilbert scan, and binary scan. Hilbert and binary scan are found to outperform raster scan. The methods considered for compression are integer wavelet transforms, a piecewise approximation with triggers (PAT) procedure introduced by Walach and Karnin (1986), and a modified PAT algorithm (MPAT). Wavelets outperform MPAT in compression, and MPAT in turn outperforms PAT; however, MPAT and PAT are computationally far simpler than wavelet transforms.