带噪图像的多扫描阈值分割编码算法的研究

ON MULTI-SCANNING THRESHOLD SEGMENTATION CODING ALGORITHM FOR NOISY IMAGE

零树编码算法是一种有效的图像编码算法,但是噪声会破坏零树结构特性,影响零树编码算法的效率,针对噪声图像提出一种基于多扫描阈值的图像分割编码算法,该算法利用多扫描阈值结构对噪声图像进行软阈值去噪、并完善逐次逼近量化过程;对重要高频子带采用图像分割编码,只对重要系数进行编码,将大量非重要系数集中成图像块不予编码,更有效地降低码率。实验结果表明,算法有效地去除了噪声,提高了编码图像的质量和编码效率,在相同的压缩比条件下,算法在编码速度、图像复原质量方面都优于EZW算法。

The zerotree coding is an efficient image compressing methods, but the noise can impact the structure character of zerotree and affect the efficiency of the coding algorithm. An image segmentation coding algorithm based on multi-scanning threshold is presented in this paper aiming at the noisy image. The algorithm uses multiple scan threshold structure to denoise the noisy image in soft-threshold way, and improves the successive approximation quantization process. The image segmentation coding algorithm is used on the important high-frequency suhbands, only the important coefficients will be encoded hut integrates large quantity of the unimportant coefficients into image block and not encodes them, thus the code rate is reduced effectively. The experiment results show that the new image compression algorithm can effectively eliminate noise, improve coding image quality and coding efficiency. Under the condition of same compression ratio, this algorithm performs better than that of EZW in aspects of recovery image quality and encoding time.