利用轮廓分形区别人工目标与自然物体算法研究

Recognition the Manmade Target and Natural Object with the Fractal Based on Contour

摘要: 根据人工目标和自然物体的固有差异，提出了一种在轮廓链码基础上提取轮廓粗糙性曲线的算法。在实际得到的水杯，车辆及岩石图像中，利用Ｖａｒｉａｔｉｏｎ算法进行了计算物体轮廓粗糙性曲线分维的实验。实验表明，水杯及车辆轮廓的分形维数较低，岩石则具有较高的分形维数，通过同二维结果的对比，从计算量的分形维数两证明用所提出的算法计算轮廓分形维数区别人工目标和自然物体是可行的。

Abstract: According to the innate difference between manmade targets and natural objects,an algorithm to abstract the rough profile of contour was presented.The variation algorithm was used to calculate the fractal dimension (FD) of the rough profile of contour of the actual image.Compared with the two dimension FD, the methos are fast and simple,the experiment result shows it is feasible to classify manmade targets and natural objects by the FD of the rough profile of contour.