基于DaVinci技术数字对焦重构算法研究和实现.docVIP

基于DaVinci技术数字对焦重构算法研究和实现 　　摘要：数字对焦光场成像技术是一种新型的计算成像技术，利用光学手段获取四维光场信息，通过重构二维空间信息和一维深度信息，达到先拍摄后对焦的目的。研究光场图像数字对焦重构算法，利用TI公司推出的DaVinci系列处理器DM6467对数字对焦重构算法进行硬件仿真，通过改变对焦系数，得到不同深度的对焦图像。同时，根据C64x+DSP核的特点，通过循环分解、项目级优化等方法对代码进行一定的优化，使代码执行效率提高了18倍左右，对实现光场成像系统的小型化和实时化具有重要的指导意义。 　　关键词：数字对焦； 图像重构； DM6467； 算法优化 　　中图分类号：TN919.85?34文献标识码：A文章编号：1004?373X(2012)04?0091?04 　　 　　Implementation of digital refocusing reconstruction algorithm based on DaVinci technology 　　PENG Ding?guo1, YUAN Yan1, ZHOU Zhi?liang2, YANG Wang?lin1 　　(1. MOE Key Laboratory of Precision Opto?mechatronics Technology, Beijing University of Aeronautics and Astronautics, Beijing 100191,China; 　　2. China University of Science and Technology, Hefei 230026, China) 　　 　　Abstract： The digital refocusing light?field photography is a new kind of computational photography, which capturs four?dimensional light?field information by optical methods and reconstructs two?dimensional spatial information and one?dimensional depth information in object space to photograph focused at different depths. Based on the research of digital refocusing reconstruction algorithms, the hardware implementation of digital refocusing algorithms for reconstruction of light?field images is presented. The DM6467 processor, one of TIs DaVinci series, is utilized as the processing platform. Images at different depths are reconstructed by changing refocusing coefficient. Moreover, code optimization methods such as cyclic decomposition, project?level optimization and so on are proposed according to the C64x+dsp core, which improves the code excutive efficiency by a factor of 18. It has a instructive significance for realizing a portable and real?time light?field imaging system in the future. 　　Keywords： digital refocusing; image reconstruction; DM6467; algorithm optimization 　　 　　 　　收稿日期：2011?09?15 　　基金项目：国家自然科学基金资助项目；长江学者和创新团队发展计划资助(IRT0705)0引言 　　传统成像技术所成的图像仅仅反应了物面光波场的二维空间分布，丢失了光波场的方向信息；而光场成像技术作为一种新型的计算成像技术，可以在一次曝光中，同时记录下入射光线