视觉假体图像处理中降采样算法的优化设计与评估  

Optimum Design and Performance Evaluation of Down-sampling Algorithm for Visual Prosthesis Imaging Processing

作　　者：古陨[1] 吴开杰[1] 雷旭平[1] 吴昊[1] 柴新禹[1] 

机构地区：[1]上海交通大学生物医学工程学院,上海市200240

出　　处：《中国医疗器械杂志》2013年第1期33-36,共4页Chinese Journal of Medical Instrumentation

基　　金：国家重点基础研究计划(973项目:2011CB707503);国家自然科学基金(91120304;60901026);高技术研究开发计划(863项目:2009AA04Z326)

摘　　要：对视觉假体图像处理系统的实时性分析表明,降采样环节的运算量是影响图像处理实时性的重要因素。为此,提出了基于间隔抽样均值的降采样算法,并以区域均值降采样算法为参照,统计分析了间隔抽样参数对图像处理效果和算法复杂度的影响。研究表明,在处理效果无显著差异的前提下,间隔抽样均值算法可以使降采样环节的运算量降低90%以上。According to analyses of the real-time performance of the visual prosthesis image processing system, down-sampling is a key factor which influences the processing speed. Based on these analyses, the algorithm of grid sampled averaging was proposed for down-sampling. Then the effect of grid parameters on the result and algorithm complexity was evaluated by using the regional-averaging algorithm as a reference. Our research shows that the grid-averaging algorithm could reduce the computation burden by 90% or more, with no significant difference from the regional-averaging algorithm.

关 键 词：视觉假体 图像处理 降采样 实时处理 

分 类 号：R318.18[医药卫生—生物医学工程]