摘要
传统的线特征提取方法在对提取生态景观的线特征时,缺少对生态景观图像进行变换处理过程,导致提取时间较长、提取结果均方误差及相对误差较大。于是研究设计了基于脊波变换的生态景观线特征提取方法。通过脊波变换对生态景观图像进行处理,达到较好的降噪效果,从而获取具有直线和超平面奇异性的信号数据。再结合SIFT法对生态景观线特征展开提取,生成尺度空间并形成特征向量。实验结果表明:上述方法的线特征提取耗时较短,且提取结果的均方误差及相对误差较小,证明该方法取得了较为理想的提取结果。
Obviously,the traditional line feature extraction method has a long extraction time,large mean square error value and relative error value during the extraction of ecological landscape features owing to the lack of transformation process.Therefore,we propose a feature extraction method of ecological landscape line based on ridge wave transform.Firstly,ridge wave transform was applied to deal with ecological landscape image for achieving excellent noise reduction effect.And then signal data with linear and hyperplane singularity were obtained.Secondly,sift method was introduced to extract the line features of the ecological landscape.Finally,the scale space was generated,forming the feature vector.The results show that the method has short time-consuming line feature extraction,minor mean square error and relative error,implying that the method has outstanding extraction results.
作者
洪堂安
杨斌
HONG Tang-an;YANG Bin(Nanchang Hangkong University,Nanchang Jiangxi 330063,China)
出处
《计算机仿真》
北大核心
2021年第8期185-189,共5页
Computer Simulation
关键词
脊波变换
尺度不变特征变换
线特征提取
图像去噪
生态景观特征
Ridgelet transform
Scale-invariant feature transform
Line feature extraction
Image denoising
Ecological landscape features
