对数似然图像分割的快速主动轮廓跟踪算法

针对跟踪目标尺度变化问题，提出了基于灰度对数似然图像分割的快速主动轮廓跟踪算法。改进的主动轮廓跟踪算法将根据以目标与背景的颜色差异而建立的对数似然图对图像进行阈值分割和数学形态学处理，再将Kalman滤波器结合到主动轮廓跟踪算法进行目标跟踪。改进的主动轮廓跟踪算法对目标分割准确，轮廓特征显著，跟踪效果稳定，算法能很好地适应跟踪目标尺度变化。通过Kalman滤波器对目标位置点的预测减少了主动轮廓跟踪算法收敛的迭代次数，使算法的运算效率提高了33%左右。

Fast active contour tracking algorithm based on log-likelihood image segmentation

A fast active contour tracking(ACT) algorithm based on log-likelihood image segmentation has been proposed to solve the scale change problem in the process of target tracking. The algorithm adopts the log-likelihood image segmentation method, which segments images according to their log-likelihood images built based on the color difference between target and background, and the mathematical morphology method, and tracks the target with conventional ACT algorithm combined with Kalman filter. It tracks the target precisely with distinct contour features and stable tracking performance, and can well adapt to the target scale change. The Kalman filter adopted reduces the number of iterations for algorithm convergence through its forecast of the target position, and thus the fast ACT algorithm is about 33% more efficient than the conventional one.