基于光流场的视频对象分割算法  

Video Object Segmentation Algorithm Based on Optical Flow Field

作　　者：葛晓叶[1] 乔栋[1] 

机构地区：[1]山西大同大学煤炭工程学院,山西大同037003

出　　处：《现代工业经济和信息化》2016年第6期86-88,共3页Modern Industrial Economy and Informationization

摘　　要：提出了一种改进的基于光流场的视频对象分割算法,先用细胞神经网络边缘叠加模板将边缘检测操作所得到的边缘与原灰度图像进行叠加,完成对原灰度图像的增强处理,再对运动目标做运动场估算,最后根据其运动特征实现视频对象的分割。本算法在保证分割结果的精确度的同时,减少了模板的使用,更加简洁,处理速度更快。This paper presents a improved video object segmentation algorithm based on optical flow field. The edge of the edge detection operation is obtained by using the edge of the cellular neural network. The superposition of it with the original gray image overlay realizes the original gray image enhancement processing. The motion field estimation of moving targets is conducted, finally video object segmentation is realized by sport characteristics. This algorithm guarantees the accuracy of the segmentation results, at the same time, reduces the use of templates, which is more concise with faster processing speed.

关 键 词：视频对象分割 光流场 细胞神经网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]