欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对现有算法存在计算量大、对摄像机和物体运动敏感等缺点,提出一种基于压缩传感（CS）和加权主色（WMC）的镜头边界检测（SBD）算法（CSWMC）。首先通过直方图特征得到粗略的镜头边界集合;然后利用CS将该集合中的帧及其前后帧的高维特征投影到低维空间,采用调节余弦相似度得到基于夹角的第一判定指标;继而定义一种新的图像主色权值和基于该权值的类Bhattacharyya相似度,得到基于颜色相似度的第二判定指标;以两种判定指标的乘积作为最终判定指标,并设计一种朴素但有效的策略进行SBD。实验结果表明,与常用方法相比,所提算法具有更高的查全率和精确率,能够更加有效进行SBD。
英文摘要：
Shot boundary detection （SBD） is regarded as one of the most critical issues in content-based video retrieval technology. The existing algorithms are sensitive to the motions caused by cameras and objects,and also need huge amount of computation. With regard to these disadvantages,a shot boundary detection algorithm based on compressive sensing and weighted major color （CSWMC） is proposed. Firstly, a coarse set of shot boundaries is obtained by comparison of frames＇ histogram features, then the high dimensional features of the frames in the set and their neighboring frames are projected to low dimensional space,and after that,the first decision criterion based on angle similarity is generated by the computation of adjusted cosine similarity. Secondly, a novel weighted value for image major color is de fined and applied to the calculation of a new Bhattacharyya-like similarity, which is used as the second decision criterion based on color similarity. Finally, the product of the two decision criteria is computed as the ultimate one. At the same time,a naive yet effective strategy is designed for boundary detection. The experimental results demonstrate that the proposed algorithm has higher recall rate and precision rate, and could perform shot boundary detection more effectively than those state of-the-art methods.
同期刊论文项目
同项目期刊论文
Universal blind image quality assessment using contourlet transform and singular-value decomposition
期刊信息
位置：