摘要： 提出一种基于多尺度自回归模型与二维TSV熵方法的红外图像分割方法。引入多尺度自回归模型，建立层与层之间及相邻层像素点之间的数学关系，并将该模型与改进的二维TSV熵最大熵方法结合，实现更合理的红外图像多尺度分割。根据相邻尺度的依赖关系，使用多尺度自回归模型的预测结果引导二维TSV熵方法对精细尺度图像进行分割，以减少最细尺度下的分割时间，去除最细尺度下的误分类斑块。实验结果表明，该方法能分割出更清晰、平滑的目标边界。

关键词: 自回归模型, 红外图像分割, 二维直方图, 二维TSV熵, 参数估计

Abstract: A method of Infrared image segmentation based on Multiscale Auto Regressive Analysis(MARA) and Two-dimensional TSV(TDTSV) entropy is proposed. MAR model is used to establish mathematic relationship among different image layers, and is combined with the method of two-dimensional TSV. Through the optimization of the regression coefficient, the adjacent scale image is combined reasonable, then it leads to a more accurate segmentation. Experimental results on infrared image show that this method reduces the iterative times of segmentation and inaccuracy classify blocks, and gets clear and smooth object border.

Key words: auto regressive model, infrared image segmentation, two-dimension histogram, two-dimension TSV entropy, parameter estimation

中图分类号: 

• TP391.41