基于图像灰度的点特征提取算子参数自适应研究

Parameters Adaptive Study of Interest Point Detect Operators Based on Image's Gray

目前基于图像灰度的点特征提取算子的参数选择均根据经验值手动设置。然而该方法常需多次设置比较后才能得到较好的特征点,所以有必要研究各算子的参数自适应。通过真实拍摄影像测试了各算子参数对提取特征点的影响情况;在此基础上提出各算子的自动设置参数算法,并通过3种对比度不同的影像分别测试了各自动设置参数算法的特征点提取效果。研究结果表明:各自动设置参数算法都能自动检测较优的参数值,避免了手动设置参数的不确定性,提高了影像特征点提取的效率。

At present the parameters of the method based on image＇ s gray to detect the interest points are setted manually based on experience, however, it often can＇t detect better interest points until compare the detection the different parameters several times. It is necessary to study the parameters adaptive of each operator. The influence of operators＇ parameters were evaluated by testing a real taken image here, on this basis, automatic parameterization algorithms of the operators were proposed, and the extraction of the algorithms which auto- matically set parameters were tested by three different contrast images. The study results showed that the al- gorithms which auto-set parameters can automatically detect the excellent parameters, avoid the uncertainty of manual-set parameters, and improve the efficiency of interest point detection.