摘要
为了减少图像稀疏分解的计算量,提出了一种基于量子遗传算法与匹配追踪相结合的图像稀疏分解快速算法.量子遗传算法能用较小的种群规模实现较大的空间搜索,全局寻优能力强,基于匹配追踪的图像稀疏分解是最优化问题,因此可用量子遗传算法快速实现.仿真结果表明,每步分解所需计算的图像或图像残差与原子的内积仅4 000次,由分解结果重建的图像具有较好的主观质量.
Based on the quantum genetic algorithm(QGA) and the matching pursuit (MP), a fast image sparse decomposition algorithm was put forward to reduce the amount of calculation. QGA combining the genetic algorithm and the quantum information theory has a large search space with small population and a good global search capability, while image sparse decomposition based on MP is an optimal problem, so it can be fast solved by QGA. Simulation results show that the number of inner product between the image or its residual image and atoms is only 4 000 times in each calculaltion step, and the reconstructed image has fine visual quality.
出处
《西南交通大学学报》
EI
CSCD
北大核心
2007年第1期19-23,共5页
Journal of Southwest Jiaotong University
基金
四川省重点科技计划项目(04GG021-020-5
03GG006-005-2)
教育部留学回国人员科研启动基金资助项目(教外司[2004]527号)
关键词
图像处理
稀疏分解
匹配追踪
量子遗传算法
image processing
sparse decomposition
matching pursuit
quantum genetic algorithm