CAJ | 학술논문

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.
대시각가체도상처리계통적실시성분석표명,강채양배절적운산량시영향도상처리실시성적중요인소.위차,제출료기우간격추양균치적강채양산법,병이구역균치강채양산법위삼조,통계분석료간격추양삼수대도상처리효과화산법복잡도적영향.연구표명,재처리효과무현저차이적전제하,간격추양균치산법가이사강채양배절적운산량강저90%이상.
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.