融合改进直方图PDE和二维Tsallis熵多阈值SAR分割

针对合成孔径雷达(SAR)图像中存在大量的相干斑噪声，对SAR图像进行分割易出现分割不精、边缘模糊等问题，融合改进的直方图PDE和二维Tsallis熵多阈值，提出了一种SAR图像分割算法。根据PDE直方图均衡化方法，将图像去噪与图像增强加权融合，利用各自权值调整去噪项与图像增强项；同时将二维Tsallis熵单阈值分割方法扩展到多阈值分割, 建立基于多阈值的选取方法，并引入萤火虫算法来求解最优阈值对，实现了二维Tsallis熵多阈值对去噪增强SAR图像的有效分割。仿真结果表明:与其他3种分割算法相比，该文算法在处理噪声大、灰度差值小的图像时具有较高的分割精度，PRI至少提升2.53%、VOI降低8.48%、GCE降低11.14%。

Multi-threshold SAR segmentation based on improved histogram PDE and two-dimensional Tsallis entropy

Due to the influence of synthetic aperture radar (SAR) imaging, there are a lot of speckle noises in the SAR image. The problem of segmentation and blurred edge is easy to appear in the segmentation of the SAR image. This paper presents a SAR image segmentation algorithm, which combines the improved histogram partial differential equation (PDE) and the two-dimensional Tsallis entropy multi threshold.The algorithm firstly combines the image denoising and the image enhancement weighting according to the PDE histogram equalization method, and uses the respective weights to adjust the denoising term and the image enhancement term to increase the image texture while denoising; the algorithm extends the 2D Tsallis entropy single threshold segmentation method to multi-threshold segmentation simultaneously, and establishes a selection method based on multiple thresholds; the algorithm also introduces the firefly algorithm to solve the optimal threshold pairs, and realizes the effective segmentation of denoising and enhancement SAR images with multi-threshold of two-dimensional Tsallis entropy.The simulation results show that compared with the other three segmentation algorithms, the proposed algorithm achieves higher segmentation accuracy when dealing with images with large noise and small gray-scale difference, which increases PRI by at least 2.53% and VOI by at least 8.48%, and reduced GCE by at least 11.14%.