进化型硬件在有损图像压缩中的应用研究  

Study on Applications of Evolutional Hardware in Lossy Image Compression

作　　者：乔双[1] 宋建中[2] 

机构地区：[1]东北师范大学物理系,长春130024 [2]中国科学院长春光学精密机械与物理研究所,长春130022

出　　处：《电子器件》2004年第1期112-115,共4页Chinese Journal of Electron Devices

基　　金：中科院长春光学精密机械与物理研究所青年创新基金(q01h01)

摘　　要：提出了一种适用于图像压缩的进化型硬件的结构,通过仿真实验充分说明了其用于图像压缩的效果优于以往的方法。在压缩过程中以函数进化的方式,使该进化型硬件能够根据图像的特点来探索最优预测函数。由于进化单元和最优函数单元是并行工作的,而且遗传操作的大部分都由硬件完成,故压缩速度很高。仿真实验表明,当PSNR=24～30 dB时,压缩比优于JPEG、Fractal、Wavelet,最高可达201:1。A structure of evolvable hardware applied to image compression is put forward, and the simulation results fully verify that its effect is better than the previous. In the process of the image compression, the use of evolution - function makes this evolvable hardware be able to search for the best prediction function. Because the evolution unit and the best - function unit are parallel while they are working, and most of the genetic operation is completed by hardware, the speed of compression is very high. The simulation results show that the compression ratio is better than JPEG, Fractal, Wavelet and maximum can attain 201:1 when PSNR = 24-30 dB.

关 键 词：进化型硬件 遗传算法 图像压缩 非线性预测 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]