Studies of Iterated Transform Image Compression and Its Application to Color and DTED

Abstract : A new classification technique based on archetypes can reduce encoding time and still maintain compression and fidelity. The use of different metrics has shown little differences in the final encoding quality thus implying that the use of the root mean square metric is preferred. Also presented are Digital Terrain Elevation Database (DTED) that are more accurate after using an iterated transform algorithm modified to encode the coastlines. The results compare favorably with an adaptive discrete cosine transformation (ADCT) and mean residual vector quantization algorithm. Additionally, results of two iterated transform algorithms for encoding 24-bit/pixel color images are presented. The first method transforms the red, green, blue (RGB) to the luminancechrominance representation before encoding, showing results similar to ADCT algorithms. The second method encodes RGB subimages directly using common transformations, and does not perform as consistently as the luminance- chrominance method. Digital Terrain Elevation Database (DTED), Archetypes, Image compression, Iterated transforms, Fractals.