计算机应用与软件
計算機應用與軟件
계산궤응용여연건
Computer Applications and Software
2015年
8期
162-165
,共4页
杨昆%张明新%刘永俊%郑金龙
楊昆%張明新%劉永俊%鄭金龍
양곤%장명신%류영준%정금룡
模板匹配%粒子群算法%归一化互相关算法%附属粒子群%黑名单机制%随机扰动算子
模闆匹配%粒子群算法%歸一化互相關算法%附屬粒子群%黑名單機製%隨機擾動算子
모판필배%입자군산법%귀일화호상관산법%부속입자군%흑명단궤제%수궤우동산자
Template matching Particle swarm optimisation (PSO)%Normalised cross-correlation (NCC)%Affiliated particle swarm%Blacklist mechanism%Random disturbance operator
针对灰度模板匹配中速度慢、抗噪性差的缺陷,基于NCC( Normalized Cross-Correlation)算法,提出一种基于优化粒子群的模板匹配算法———NPSO。该算法加入附属粒子群,引导主粒子群向全局最优解收敛;根据禁忌搜索思想,提出黑名单概念,使粒子群快速跳出局部最优;并引入随机扰动算子,增加粒子群向全局最优解收敛准确性。通过Matlab仿真实验,不同模板尺寸下NPSO精确匹配率比基于标准粒子群模板匹配算法分别提高了45%、79%、36%、2%,且对噪声不敏感。说明NPSO不容易陷入局部最优,且匹配精度高、抗噪能力强。
針對灰度模闆匹配中速度慢、抗譟性差的缺陷,基于NCC( Normalized Cross-Correlation)算法,提齣一種基于優化粒子群的模闆匹配算法———NPSO。該算法加入附屬粒子群,引導主粒子群嚮全跼最優解收斂;根據禁忌搜索思想,提齣黑名單概唸,使粒子群快速跳齣跼部最優;併引入隨機擾動算子,增加粒子群嚮全跼最優解收斂準確性。通過Matlab倣真實驗,不同模闆呎吋下NPSO精確匹配率比基于標準粒子群模闆匹配算法分彆提高瞭45%、79%、36%、2%,且對譟聲不敏感。說明NPSO不容易陷入跼部最優,且匹配精度高、抗譟能力彊。
침대회도모판필배중속도만、항조성차적결함,기우NCC( Normalized Cross-Correlation)산법,제출일충기우우화입자군적모판필배산법———NPSO。해산법가입부속입자군,인도주입자군향전국최우해수렴;근거금기수색사상,제출흑명단개념,사입자군쾌속도출국부최우;병인입수궤우동산자,증가입자군향전국최우해수렴준학성。통과Matlab방진실험,불동모판척촌하NPSO정학필배솔비기우표준입자군모판필배산법분별제고료45%、79%、36%、2%,차대조성불민감。설명NPSO불용역함입국부최우,차필배정도고、항조능력강。
Template matching based on grayscale has drawbacks of running slow and poor noise immunity.Aiming at the problems and based on NCC algorithm, we propose a PSO-based template matching method named NPSO.The method adds the affiliated particle swarm and leads the primary particle swarm to converge to the global optimum solution.According to the idea of tabu search, we put forward the concept of the blacklist to make the particle swarm jump out of the local optima quickly.Moreover, we introduce random disturbance operator which increases the accuracy of particle swarm convergence to the global optimum solution.It is shown by the Matlab simulation experiment that the precise matching rates of NPSO with different sizes of template increase 45%, 79%, 36%and 2%respectively than the template matching algorithm based on standard particle swarm, and they are insensitive to noise.This demonstrates that NPSO is not easy to fall into local optimum, the matching accuracy is high, and the anti-noise capability is strong as well.