In block truncation coding (BTC), an image is segmented into n x n (typically, 4 x 4) nonoverlapping blocks of pixels, and a two-level (one-bit) quantizer is independently designed for each block. Both the quantizer threshold and the two reconstruction levels are varied in response to the local statistics of a block. Thus, encoding is essentially a local binarization process, and the representation of a block consists of an n x n bit map indicating the reconstruction level associated with each pixel and overhead information specifying the two reconstruction levels. Decoding is the simple process of placing the appropriate reconstruction value at each pixel location as per the bit map. A diagram of the basic BTC scheme is shown in Fig. 11.1.
Online access to SPIE eBooks is limited to subscribing institutions.