计算机工程与应用 ›› 2014, Vol. 50 ›› Issue (21): 171-174.

• 图形图像处理 • 上一篇    下一篇

基于积分图和粒子群优化的肤色分割

朱志亮1,刘富国1,陶向阳1,2,刘晓山1,2   

  1. 1.江西师范大学 物理与通信电子学院,南昌 330022
    2.江西省光电子与通信重点实验室,南昌 330022
  • 出版日期:2014-11-01 发布日期:2014-10-28

Skin segmentation based on integral image and particle swarm optimization

ZHU Zhiliang1, LIU Fuguo1, TAO Xiangyang1,2, LIU Xiaoshan1,2   

  1. 1.College of Physics, Communications and Electronics, Jiangxi Normal University, Nanchang 330022, China
    2.Key Laboratory of Photoelectronic & Telecommunication of Jiangxi Province, Nanchang 330022, China
  • Online:2014-11-01 Published:2014-10-28

摘要: 为了更快地将肤色区域从图像中分割出来,在二维Ostu方法的基础上,提出了一种将积分图和粒子群优化相结合用于寻找最优分割阈值点的新方法。该方法通过对经过颜色补偿后的图片在YCbCr颜色空间中用二维高斯肤色模型计算肤色概率图,根据二维Ostu算法确定粒子群优化的目标函数,将积分图的方法用于粒子群优化中目标函数的计算,在有效地获取最佳阈值点的同时减少了计算量。实验表明,该方法能有效并且快速地分割出肤色区域。

关键词: 肤色分割, 高斯肤色模型, Ostu算法, 积分图, 粒子群优化

Abstract: To segment the skin color area faster, this paper puts forward a method to look for the best segmentation threshold point with integral image and particle swarm optimization based on the basic 2D Ostu method. The method uses two-dimensional gaussian skin model in the YCbCr color space with a color compensation picture to gain the skin probability graph, and then confirms the particle swarm optimization’s objective function by the 2D Ostu algorithm, and calculates the function’s value with the integral image. The method can get the best threshold point effectively and can shorten the calculation at the same time. Experiments show that the method can segment the skin color area effectively and quickly.

Key words: skin segmentation, gaussian skin model, Ostu algorithm, integral image, particle swarm optimization