改进的多粒度原理在遥感图像分割算法研究

Remote Sensing Image Segmentation Algorithm Based on an Improved Multi Granularity Principle

针对传统的遥感图像分割算法由于计算复杂等原因,造成图像的分割分辨率低,清晰度不高,当图像中的信息量非常大时,对图像分割非常耗时等问题缺陷,为了有效地分割图像,提出了一种改进的多粒度原理和小波算法相结合的遥感图像分割算法;该方法首先采用小波变换对图像的弧度直方图进行小波多尺度变换,并进行分解操作,然后采用粒度合成技术对分解后的图像进行合成;文中采用的是256×256的SAR图像来进行实验对比,结果表明,提出的算法有效地改善了分割效果,分割出的图像边缘效果明显清晰,证明了该算法的可行性和有效性。

In view of the traditional image segmentation of remote sensing image due to the computational complex and other reasons, resulting in low resolution image segmentation, clarity is not high at the same time, when the image information in a very large amount of image segmentation, is time consuming and defects. In order to improve image segmentation, this paper proposes an improved multi granularity theory and wavelet algorithm combining image segmentation of remote sensing image. The method firstly uses the wavelet transform to the image histogram of the radian of muhiscale wavelet transform, and then uses decomposition operation, granular synthesis technology on the decomposition of the image after synthesis. The experimental results show that the proposed algorithm effectively, improve the segmentation effect, the segmented image edge effect is distinct, proved the feasibility and effectiveness of the proposed algorithm.