计算机应用研究
計算機應用研究
계산궤응용연구
APPLICATION RESEARCH OF COMPUTERS
2010年
1期
361-364
,共4页
于天虎%毛兴鹏%王国谦%马孝阳
于天虎%毛興鵬%王國謙%馬孝暘
우천호%모흥붕%왕국겸%마효양
图像分析%图像分割%边界检测%梯度%各向同性
圖像分析%圖像分割%邊界檢測%梯度%各嚮同性
도상분석%도상분할%변계검측%제도%각향동성
image analysis%image segmentation%edge detection%gradient%isotropy
为降低基于梯度的边界检测算法的复杂度,常使用两种梯度近似算法.但这些梯度近似值受边界方向的影响较大,降低了边界检测的性能.提出了通用梯度近似算法的数学模型和两种优化准则,进而推导出两种梯度近似的优化算法.分析表明:与常用算法相比,优化算法在各向同性的性能方面提高4.4倍,在梯度幅度的逼近度提高5～7倍.同时,给出了优化算法的简单快捷的实现方法.
為降低基于梯度的邊界檢測算法的複雜度,常使用兩種梯度近似算法.但這些梯度近似值受邊界方嚮的影響較大,降低瞭邊界檢測的性能.提齣瞭通用梯度近似算法的數學模型和兩種優化準則,進而推導齣兩種梯度近似的優化算法.分析錶明:與常用算法相比,優化算法在各嚮同性的性能方麵提高4.4倍,在梯度幅度的逼近度提高5～7倍.同時,給齣瞭優化算法的簡單快捷的實現方法.
위강저기우제도적변계검측산법적복잡도,상사용량충제도근사산법.단저사제도근사치수변계방향적영향교대,강저료변계검측적성능.제출료통용제도근사산법적수학모형화량충우화준칙,진이추도출량충제도근사적우화산법.분석표명:여상용산법상비,우화산법재각향동성적성능방면제고4.4배,재제도폭도적핍근도제고5～7배.동시,급출료우화산법적간단쾌첩적실현방법.
Two approximations to gradient have been introduced to reduce the computational complexity of the gradient based edge detection algorithm. However, these approximations yield anisotropy so that the performance of the edge detection deteriorates.This paper firstly established a general mathematical model for the approximation to gradient and suggested two optimization principles of the approximation model, and then developed two optimal algorithms to reach minimum anisotropy or minimum approximation error while keeping computational simplicity.In addition,proposed some simple and direct hard-ware implementations.The mathematical analyses indicate that the performance of the proposed optimal algorithm gets better 4.4 times in isotropy and 5-7 times in approximation error than the existing approximations, and the computer simulation results support the analysis.
