基于DNA计算的图像模板匹配算法

Image template matching algorithm using DNA computation

针对已有的图像匹配算法都是在小规模基础上的统计识别方法,均是串行运算,对样本的训练和目标识别都须进行大量复杂的运算,难以适应大规模图像比对的问题,利用DNA计算强大的并行性,提出了一种基于DNA计算的图像模板匹配算法.首先,将二进制的图像信号编码为满足一系列约束并允许一定非特异性杂交的DNA序列;然后,通过DNA退火反应得到匹配问题的解;最后,利用这一算法得出的128个单链DNA编码进行了数字图像模板匹配的仿真实验.仿真结果表明DNA计算应用于大规模图像匹配问题是可行的.

A DNA （deoxyribnucleic acid） algorithm was used to solve matching problems arisen from image alignment. An algorithm was proposed to map binary value into DNA codewords by satisfying a number of constraints, including the noise tolerance constraint. Then, the matching problem could be solved with DNA annealing reaction. At last, from the algorithm, 128 codewords were retrieved to use a DNA based approach to digital image alignment. Simulation results show that the DNA computing in the field of digital signal processing will be useful.