一种有效的遥感图像无缝分割方法

遥感图像具有数据量大、模糊性较强、纹理细节丰富等特点，这就决定了遥感图像的分割与自然景色图像分割相比，在分割效率、分割效果上都提出了更高的要求。另外，无缝空间遥感影像数据库的建立为我们提出了在基于内容的影像检索中遥感图像无缝分割问题。在JSEG算法的基础上，通过图像增强技术，解决了遥感图像不同区域边界往往不明显的问题，大大改善了分割效果；为了解决分割的无缝性，提出了基于gabor纹理特征的区域边界合并算法。实践证明本方法很好地解决了遥感图像分割的无缝性问题。

An Efficient Approach for Seamless Segmentation of Remote Sensing Image

Remote sensing image has the characteristics such as huge data quantity, intense illegibility, and rich texture, which leads to higher requirement on its segmentation compared with nature scene image. In addition, the creation of seamless spacial remote sensing image database brought forward the question about the seamless segmentation of remote sensing image,which is in the field of content-based image retrieval. In this article, based on JSEG algorithm and image-strengthen technique, we solved the problem of remote sensing image that boundary of different area is often inconspicuous and greatly meliorated the segmentation effect. In order to implement the seamlessness in segmentation, an area boundary merging algorithm based on gabor texture features was proposed. Practice proved that the algorithm well implemented seamlessness of remote sensing image segmentation.'