论文部分内容阅读
提出了一种复杂背景下红外图像序列中圆形目标提取、判别以及跟踪的系统框架。在目标提取阶段,采用支持向量回归的方法选择种子点,用自适应选择阈值区域的生长方法进行分割提取目标。对于目标判别,采用最多共圆像素点数除以总像素点数作为圆形程度的度量进行圆形判别,并提出改进的标准Hough变换来找到共圆点。跟踪算法中,应用了粒子滤波的跟踪方法,并针对红外目标的特点,建立实现粒子滤波算法的细节。实验证明了整个框架体系的有效性和稳健性。
A system framework of circular object extraction, discrimination and tracking in infrared image sequences under complex background is proposed. In the target extraction stage, the support vector regression method is used to select the seed points, and the adaptive threshold selection method is used to segment and extract the target. For the purpose of discrimination, using the maximum number of common pixels divided by the total number of pixels as a measure of the circular degree of circular discrimination, and proposed improved standard Hough transform to find the common dot. In the tracking algorithm, the tracking method of particle filter is applied. For the characteristics of infrared target, the detail of particle filter algorithm is established. The experiment proves the validity and robustness of the whole framework system.