论文部分内容阅读
针对红外图像中弱小目标的检测问题,提出了一种基于图像背景纹理分析的检测方法,利用二阶高斯-马尔柯夫随机场(GMRF)模型对背景纹理进行描述,并采用最小平方误差方法进行参数估计,将估计参数作为纹理的特征向量。目标检测采取由粗到精两个步骤,首先根据特征向量将目标区域粗定位,然后利用估计出的参数对目标可能存在的区域进行模型拟合,根据拟合误差检测出目标位置。实验结果证明了这种方法的有效性。
Aiming at the detection of weak targets in infrared images, a detection method based on image background texture analysis is proposed. The background texture is described by the second-order Gaussian-Markov random field (GMRF) model and the least square error method Parameter estimation, the estimated parameters as texture eigenvectors. The target detection takes two steps: rough to fine. Firstly, the target area is coarsely located according to the eigenvector. Then, the estimated parameters are used to fit the target area and the target position is detected according to the fitting error. Experimental results show the effectiveness of this method.