计算机工程与设计
計算機工程與設計
계산궤공정여설계
COMPUTER ENGINEERING AND DESIGN
2015年
8期
2129-2132,2142
,共5页
尺度不变特征转换%小波变换%高斯金字塔%降采样%随机抽样一致
呎度不變特徵轉換%小波變換%高斯金字塔%降採樣%隨機抽樣一緻
척도불변특정전환%소파변환%고사금자탑%강채양%수궤추양일치
scale-invariant feature transform%wavelet transform%Gaussian pyramid%down-sampling%random-sample consensus
为解决尺度不变特征转换匹配算法运行时间过长、匹配率不高的问题，提出一种改进的尺度不变特征匹配算法。在原经典的尺度不变特征转换匹配算法的基础上，引入二维 Mallat快速小波变换算法，重建图像的低频成分；对高斯金字塔组数进行调整，减少降采样次数；通过优化的随机抽样一致算法剔除误匹配点。MATLAB 仿真结果表明，改进后的算法减少了匹配耗时，提高了匹配率，优于原算法。
為解決呎度不變特徵轉換匹配算法運行時間過長、匹配率不高的問題，提齣一種改進的呎度不變特徵匹配算法。在原經典的呎度不變特徵轉換匹配算法的基礎上，引入二維 Mallat快速小波變換算法，重建圖像的低頻成分；對高斯金字塔組數進行調整，減少降採樣次數；通過優化的隨機抽樣一緻算法剔除誤匹配點。MATLAB 倣真結果錶明，改進後的算法減少瞭匹配耗時，提高瞭匹配率，優于原算法。
위해결척도불변특정전환필배산법운행시간과장、필배솔불고적문제，제출일충개진적척도불변특정필배산법。재원경전적척도불변특정전환필배산법적기출상，인입이유 Mallat쾌속소파변환산법，중건도상적저빈성분；대고사금자탑조수진행조정，감소강채양차수；통과우화적수궤추양일치산법척제오필배점。MATLAB 방진결과표명，개진후적산법감소료필배모시，제고료필배솔，우우원산법。
To solve the problems of scale-invariant feature transform algorithm including the long running time and the low match rate,an improved SIFT algorithm was proposed.On the basis of the original classic scale-invariant feature transform algorithm, a two-dimensional Mallat fast wavelet transform algorithm was introduced,low-frequency components of the image were recon-structed.The number of the Gaussian pyramid group was revised,and the number of down-sampling was reduced.The mis-matching points were finally removed using improved random-sample consensus algorithm.MATLAB simulation results show that the improved algorithm not only reduces the matching time costs,but also improves the match rate,which is better than the original algorithm.
