摘要:
基于内容的视频信号(Video)与图像库检索是当前计算机视觉,图像数据库与知识
挖掘(Knowledge Discovery)等领域研究的热点之一.较系统地介绍了该研究方向的现状.对
于静态图像,介绍了基于颜色、纹理、形状、区域、目标特征的检索和交互式检索方法,对于视
频序列图像,介绍了镜头检测,镜头内容表示,场景建立,运动目标分割等技术.最后分析了该
领域研究存在的难点,并提出了对今后工作的展望.
Abstract:
Content-based image/video parsing and retrieving is one of most active
researches in computer vision, image database, knowledge discovery and so on. In
this paper, we systemically introduce the technical development situation in this
area. For image, we introduce the techniques based on color, texture, shape, region,
object feature and interactive retrieval method. For video, we discuss the techniques
for shot detection ,shot content representation ,scene construction and moving object
segmentation. Finally, we analysis the difficulties in this area and make some
remarks about future work.