基于交叉熵及曲线进化的图像分割算法

该文提出一种新的图像分割算法,从目标和背景区域的差异性出发,利用信息论中的交叉熵作为衡量标准,构造能量函数,通过最小化能量即可得到分割结果。在最小化能量函数时,运用最陡梯度下降法导出曲线进化方程,然后考虑噪声的影响提出了改进模型,并用水平集方法来表示此曲线进化方程,利用快速水平集方法来进行数值求解。最后的仿真结果证明了本文算法的有效性。

An Image Segmentation Algorithm Based on Cross Entropy and Curve Evolution

A novel algorithm for image segmentation is presented.We take into account the dissimilarity between object and background in image and utilize the cross entropy as measure criterion.The criterion is formulated as an energy function.The energy function is minimized by using gradient-descent methods,which leads to a curve evolution equation that segments the image.Considering the affect of noise,we present an improved model,and use level set method to represent the curve evolution equation,and the equation is solved by utilizing fast level set method.The experimental results show that the proposed algorithm is efficient.