基于主体外部轮廓的照片分割算法

在实际应用中常常需要将人像从照片的背景中有效地分割出来，通常的做法是通过提取照片中人像各部分特征来实现分割，其缺陷是计算复杂度较大，分割效果受各种干扰因素影响很大。现提出了一种基于主体外部轮廓信息的照片分割算法，通过梯度锐化、边缘保护、临近像素点相似度比较来确定图像的边界，该算法对于具有相对均匀背景的照片具有较高的计算效率。

Segmentation algorithm for photos based on outer outline of the main body

It is often needed to separate human's image from a photo effectively in actual work.The conventional algorithm used is based on extracting characteristics of the main body in the picture,the defects of which are the high complexity of computation and the result is easily affected by much noises.In this paper a new algorithm is presented based on the outer outline of the human's image in the photo,which includes gradient sharpening,edge protection and comparing the similarity with the pixels nearby.When applying the algorithm to the photo with relatively even background,it shows high efficiency of computation.