论文部分内容阅读
图像分割是面向对象图像分析的基础。目前常规的图像分割算法普遍基于光谱同质性假设,但是这种假设对于提取干旱地区盐田这种具有析出结晶盐与卤水两种高反差地物共存的空间对象而言显得不足为用。针对面向对象图像分析中只采用光谱和形状异质进行图像分割的不足,以吉兰泰盐田及周边地区2008年11月SPOT 5影像为例,首先采用窗口傅立叶变换功率谱方法提取影像纹理特征,然后进行基于纹理、光谱的多尺度分割,进而对分割后图像进行多层次分类来提取盐田信息。实验结果表明,该方法对盐田地区的信息提取有较好的效果。
Image segmentation is the basis of object-oriented image analysis. At present, the conventional image segmentation algorithms are generally based on the assumption of spectral homogeneity, but this assumption is not sufficient for the extraction of space objects in the salt fields of arid areas, which have two kinds of high contrast objects such as precipitated crystal salt and brine. Aiming at the shortcomings of image segmentation using only spectral and shape heterogeneity in object-oriented image analysis, taking image of SPOT 5 in Yitian and surrounding areas of Jitlantai for example, the image texture feature is first extracted by window Fourier transform power spectrum method, Then, the multi-scale segmentation based on the texture and the spectrum is performed, and the multi-level classification of the segmented image is performed to extract the salt information. Experimental results show that this method has a good effect on extracting information from Yantian area.