Abstract:

The traditional high order quantity error correction codes, based on the tent chaotic map, have the feature of linear error correction to image. And the “threshold effect” of compressed image peak signal-to-noise ratio (PSNR) can be resolved. And then the elegant feature and theory image PSNR limit of the error correction codes, based on the tent map, are analyzed. We can get that the accurate estimation of sign bit is the key to improving performance. However, the traditional harddecision decoding of sign bit becomes the bottleneck to restrict the improvement of performance. That is because the sign bit can not be protected effectively by the traditional high order quantity codes. Based on this, the high order quantity codes, based on softdecision decoding of sign bit, are proposed. In addition, combined with low density parity check (LDPC) codes, the sign bit can be protected effectively with the softdecision decoding method. Finally, the simulation results show that the image PSNR can approach the theory limit using the proposed high order quantity codes. And compared with the traditional high order quantity codes, the image PSNR can get the gain of 2 dB with the almost same encoding and decoding complexity, when the initial probability of sign bit is obtained by looking up tables.