一种融合对数极坐标映射和归一化割的图像分割算法

Image segmentation approach by combining log-polar transformation and normalized cut

为了缩减归一化割算法中权值矩阵的规模,提出了一种融合对数极坐标和归一化割的图像分割算法。首先将均匀采样的图像映射到对数极坐标系下,然后进行归一化割运算,达到减少时间复杂度、提高算法执行效率的目的。实验结果表明该方法能够取得理想的分割效果。

In order to curtail the weight matrix of normalized cut,this paper proposed an image segmentation algorithm based on LPT and normalized cut.First,it mapped the uniform sampling image from Cartesian to log-polar coordinate.Then,it implemented the image segmentation based on the normalized cut method.Consequently,it achieved the purpose of reducing time complexity and improving computational efficiency.Experimental results demonstrate that the proposed method can obtain desired segmentation results.