可变尺度区域拟合模型的侧扫声纳分割方法

为解决现有侧扫声纳图像目标分割准确度不高的问题,提出一种联合最大熵去噪和可变尺度区域拟合模型的侧扫声纳图像分割方法。首先,计算图像一维熵,基于最大熵原则对侧扫图像进行降噪处理,提高图像质量,并根据峰值信噪比评判降噪效果;然后基于可变尺度区域拟合模型,采用高斯核函数对分割活动轮廓进行约束,分割降噪后的侧扫声纳图像。通过对含有不同目标物的侧扫声纳图像进行分割实验,验证了联合最大熵去噪和可变尺度区域拟合模型的有效性。

Side scan sonar image segmentation method of region-scalable fitting energy model

In order to solve the low accuracy of side scan sonar image segmentation,a region-based active contour method of side scan sonar image segmentation is proposed.In order to eliminate the noise of the side scan sonar image,the 1D entropy is calculated and the principle of maximum entropy is used to reduce the noise of the side scan sonar image,and evaluate the denoising effect according to the peak signal-to-noise ratio firstly; then based on the Region-Scalable Fitting Energy model,Gaussian kernel function is used to constrain the segmentation active contour.Through the segmentation experiment of side scan sonar images containing different targets,the effectiveness of the Maximum entropy denoising and the Region-Scalable Fitting Energy model is verified.The Region-Scalable Fitting Energy proposed in this paper provides a method to segment high-noise side scan sonar images.