Research On Video Object Segmentation Technology

In recent years, multimedia technology has been rapidly developed. There is more people focus on the audio and video services, interactive service. This also had promoted the video object segmentation technology progressed. So far, there have been hundreds of segmentation algorithm and the new theory has been applied to video object segmentation. There are a lot of video object segmentation methods and there is no unified classification criterion. According to the temporal and spatial relationships, it can be classified based on time domain segmentation. Segmentation technology based on the spatial and temporal information of the segmentation technology based on three categories. This article focuses on these three aspects of video segmentation research technology.Based on intra frame (spatial) segmentation of video object technology is adopted in traditional image segmentation based on the technology and development, firstly, the spatial information of image such as color, texture, consistency, the edge of image regions are divided by partition results for object tracking in the frame for continuous segmentation. In addition to the space information, can also be used for image region using the transform domain information, a priori knowledge.Based on inter frame difference (Time domain), opticalflow, movement information detection template using segmentation technique, and then through some post-processing on detection template to obtain video object template, finally using some advanced algorithms to guide the subsequent frames of the video object segmentation.Segmentation algorithm based on spatio-temporal information can make full use of the time and spatial information based on, is a relatively promising segmentation algorithm.In additional, stereo video technology is an important development direction of multimedia technology in the future. It is a new type of video technology can provide stereo feeling. Therefore, this paper studied on stereo video objects segmentation. Stereo video series not only can use the three kinds of technical information, but also the depth information can obtain reliable segmentation for object. Based on the research of stereo video object segmentation is proposed in this paper. Stereoscopic video object segmentation based on active contour and neural network, the related experimental results show that, this method is effective.