基于改进ACFOA的图像一维OMP稀疏分解

IMAGE 1D OMP SPARSE DECOMPOSITION BASED ON MODIFIED ACFOA

针对二维图像稀疏分解运算复杂度高的问题,提出一种基于改进自适应混沌果蝇优化算法的图像一维正交匹配追踪OMP(Orthogonal Matching Pursuit)稀疏分解方法。算法首先将图像从二维空间转换到一维空间,然后对自适应混沌果蝇优化算法ACFOA(Adaptive Chaos Fruit Fly Optimisation Algorithm)的味道浓度判定值和混沌映射函数进行了改进,提高了算法的全局寻优性能,最后将改进后的ACFOA算法应用到图像一维OMP分解之中。实验结果表明,在相同实验条件下,图像一维OMP稀疏分解的速度是二维分解的1.12倍。

2D image sparse decomposition has the problem of high computational complexity. In order to solve this,we presented the image 1D orthogonal matching pursuit（ OMP） sparse decomposition algorithm, which is based on modified adaptive chaos fruit fly optimisation algorithm（ ACFOA）. First,the algorithm converts the image from 2D space to 1D space. Then it modifies the flavour concentration determination value and chaotic mapping function of the ACFOA,improves the global optimisation performance of the algorithm.Finally,we applied the improved ACFOA in image 1D OMP decomposition. Experimental results showed that the speed of image 1D OMP algorithm was 1. 12 times faster than 2D decomposition under the same condition.