【摘要】 该文提出了一种基于改进小波域隐马尔可夫树(HMT)模型进行图像分割的方法。该方法利用基于希尔伯特变换对的二维方向小波,这种小波变换具有平移不变性、方向检测性好的特点。同时该方法还利用拓展HMT对该改进小波域中尺度间的小波系数相关性进行建模,并结合多背景融合技术进行遥感图像的分割,得到了优于已有文献的分割结果,而且与同类算法相比,降低了算法所需的计算量。更多还原

【Abstract】 Improved wavelet-domain HMT based remote-sensing image segmentation algorithm is proposed in this paper. The algorithm is based on 2-D directional wavelet, which is implemented via Hilbert transform pairs. The 2-D directional wavelet can provide both shift invariance and good directional selectivity. In this paper, the dependence of wavelet coefficients lied in inter scale is modeled efficiently, and a new segmentaion algorithm is produced by combining this with multicontext fusion method. A better segmentation result for remote-sensing image with smaller computational burden is obtained.更多还原