摘要
采用概率论理论 ,对二维图像进行灰度统计分析 ,采用计算标准差方法 ,对图像特征点区域定位并提取特征点。该方法提取特征点 ,仅需对抽样象素区域进行灰度标准差分析 ,避免了提取图像特征点 ,根据被处理图像的一些先验信息 ,利用试探方法确定阈值的局限性。通过分别对具有弱纹理及包含复杂背景的多物体自然二维图像的特征提取 ,证实了所提方法的有效性和可靠性 ,可满足机器视觉系统中自主、实时识别与提取二维图像特征点要求。
The 2-D image is analyzed on gray statistically with probability theory. The area of feature points across the 2-D gray images can be located and the feature points would be extracted by calculating standard deviation on sampled pixel region.It only needs to compute standard deviation of gray within sampled pixel area for extracting feature points with our approach. It can be avoided that the limitation to determine threshold value by tentative method according to some prior information on processing image.The algorithm mentioned in the paper is valid and reliable by extracting feature points on actual natural images with weak texture and complex background including multi-object. It can meet the demands of extracting feature points of 2-D image automatically in machine vision system.
出处
《传感技术学报》
CAS
CSCD
2004年第1期70-73,共4页
Chinese Journal of Sensors and Actuators
