结合H．264压缩域和像素域运动目标分割

摘要

The moving object segmentation algorithm with H.264 compressed domain is faster in segmentation speed, while the one with pixel domain is more accurate in segmentation.In this paper we present a new segmentation method, which combines the motion information in compression domain with the texture information in pixel domain.The motion vector field is obtained in compressed domain first, after filtering, normalisation and motion vector field accumulation, a relatively reliable motion area is obtained.Then we use local binary pattern to describe texture information, and finally the accurate segmentation is made.Experimental results show that under the condition of ensuring same segmentation accuracy, the speed of segmentation is greatly improved.%H．264压缩域运动目标分割算法分割速度较快，像素域分割较为准确。提出一种结合压缩域运动信息和像素域纹理信息的分割方法。在压缩域得到运动矢量场，经过滤波、归一化和运动矢量场累加，得到相对可靠的运动区域，然后采用局部二值模式描述纹理信息，最后再进行精确分割。实验结果表明，在保证分割准确度的情况下，大大提高了分割速度。