摘要
针对航拍图像由于受噪声和光照影响导致分割精度差问题,将一种改进的C-V水平集方法引入到航拍图像运动分割中。首先对C-V模型作了改进,通过引入新的能量项使改进方法无需重新初始化符号距离函数,在保证分割精度的同时提高了分割速度;其次对水平集函数的偏微分方程进行改进,消除了原始C-V方法对远离实际边缘的轮廓曲线进化的抑制作用。实验结果显示,改进方法能有效地分割出航拍图像中的运动目标,在初始轮廓线远离目标实际轮廓时也能够正确分割,并能更好地满足实时性的要求。改进方法具有分割精度高,耗时少的优点,可用于航拍图像运动目标检测跟踪系统。
In order to improve the poor precision of aerial image segmentation caused by illumination and noise, a revised level set method is proposed. First, the paper presented a new energy item that penalized the deviation of the level set function from a signed distance function, which reduced the heavy computation in re - initialization of SDF while ensuring segmentation precision. Secondly, it improved the PDE by replacing the Dirac functionδ（φ）with ｜↓△ φ（x,y） ｜, which eliminates the restraining effects when evolution curve is far from the actual contour. Experimental results show that the improved method can effectively segment moving objects out of background from aerial images and better meet the requirements of the real - time processing. The revised method has significant predominance in terms of robust and fast performance, and it can also achieve a satisfactory result even when the initial contour is far from the real contour. With the advantages of high precision as well as low time cost, this method can be used for tracking system in videos taken from unmanned aerial vehicles.
出处
《计算机仿真》
CSCD
北大核心
2009年第2期243-246,共4页
Computer Simulation
基金
国家自然科学基金重点项目(60634030)
航空科学基金(2007ZC3037)
关键词
航拍图像
水平集
运动检测
图像分割
Aerial image
Level set
Motion detection
Image segmentation
