基于模糊ID3决策树的快速角点检测算法

Method for high-speed corner detection based on the fuzzy ID3 decision tree

为了解决现有角点检测算法普遍存在计算时间长、效率不高、不适于实时在线检测的问题,提出了一种基于模糊ID3决策树的快速角点检测算法。利用半径为3像素的Bresenham圆环作为检测模板,在被检测图像上移动,圆环中心与备选角点重合;比较备选角点与圆环上像素点亮度值的高低,应用隶属度函数做出模糊化判断;进而根据该文定义的分类规则和模糊ID3算法,选取圆环上信息增益最大的像素点作为父节点,建立二元模糊决策树,实现角点与非角点的分类。采用实地采集于北京丰台铁路段的图片对提出算法从检测精度、计算速度、抗噪性能等方面进行验证。实验结果表明:当在检测圆环上选取比较像素点的个数为9时,可以获得最佳的检测效果;在保证角点检测准确性前提下,比较目前常用的几种角点检测算法,计算效率有明显提高;同时,该文算法对Poisson噪声、Gaussian噪声都具有良好的抗噪性能。

The previous algorithms for corner detection are computationally intensive for using in real-time applications of any complexity.A high-speed corner detection algorithm was developed to solve the problem.The algorithm uses a 3 pixels diameter Bresenham＇s circle as the test mask,overlapping the candidate corners with the nucleus.Pixels on the circle are compared with the nucleus of the intensity value,with the membership function used to give the fuzzified result.The pixel with maximum information gain was then chosen as the parent node by Fuzzy ID3 algorithm,according to the classification rules defined in this paper.A binary decision tree was built to partition the corners and non-corners.The pictures taken in Fengtai Railway Station in Beijing were used to test the method.Experimental results show that when the number of pixels on the test mask chosen to be 9,best detection result can be obtained.The corner detector significantly outperforms existing detectors in computational efficiency without sacrificing the quality and the method also provides high performance against Poisson noise and Gaussian noise.