DefinePK

Abstract

Various imaging applications have adaptively used the lossless Huffman entropy coding technique since the redundancy of an image data will be expelled at the precise level. We proposed an accelerated parallel Huffman entropy coding algorithm which implements on the environment of GPGPU using CUDA architecture. This algorithm proceeds with a parallel histogram approach for determining the occurrence of every symbol of an input data, from that we generate a code word for every symbol after the construction of a sequential Huffman tree. Subsequently, the compressed data obtained as in hexadecimal form after applying the adaptive approach where it reads the data parallel as word wise (8, 16, 32, 64 bits) to the code word. The experimental results showed that the GPGPU runs the parallel code with the speed of 46X than the CPU's serial code.