A Fast Gradual Shot Boundary Detection Method Based on SURF
Video shot boundary detection is a key technology of content-based video retrieval, and has been extensively studied in these years. But researchers tend to sacrifice speed for a higher accuracy. To improve the performance of the algorithm and reduce the
- PDF / 332,102 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 22 Downloads / 240 Views
Abstract Video shot boundary detection is a key technology of content-based video retrieval, and has been extensively studied in these years. But researchers tend to sacrifice speed for a higher accuracy. To improve the performance of the algorithm and reduce the computational cost, Hue, Saturation, Value (HSV) histogram is introduced into the pre-processing of the video to output the candidate gradual shot segments. Then we use speeded up robust features (SURF) to detect the gradual shot boundaries. The experiment results show that the method can improve the speed with a high degree of accuracy. Keywords Video retrieval HSV histogram SURF
Gradual shot change Shot boundary detection
1 Introduction With the rapid development of computer network and multimedia technology, digital videos have being gotten more and more extensive applications, people are increasingly concerned about how to retrieve the information they need from a bilious stream of videos. Therefore, It has become important that how to organize and retrieve the video information, especially in the field of database and information retrieval. The information retrieval which is based on textual data originally cannot meet the demand of the people. Because of the variety of the video data, the subjectivity of human when describing video content and the consuming time of generating text indexing manually, the content-based video retrieval technology emerges, and becomes more and more popular in the field of multimedia.
Z. Wu (&) P. Xu Communication University of China, Beijing 100024, China e-mail: [email protected]
Z. Wen and T. Li (eds.), Practical Applications of Intelligent Systems, Advances in Intelligent Systems and Computing 279, DOI: 10.1007/978-3-642-54927-4_66, Springer-Verlag Berlin Heidelberg 2014
699
700
Z. Wu and P. Xu
The so-called video shot boundary detection is segmenting the video into many shots automatically and can determine the types of shot change. It is the first and most important step in content-based video retrieval. Because of the complexity of the scenes in the video, there are many approaches to implement the shot boundary detection. Such as the algorithm based on pixel comparison [1], the algorithm based on histogram [2], the algorithm based on edge features [3] and the algorithm based on compressed domain, such as the algorithm based on DCT coefficients [4] and the algorithm based on motion vectors [5]. Generally speaking, there are two kinds of shot change types, one is of shot abrupt change and the other is shot abrupt change. Yet many researchers tend to sacrifice the speed for a higher accuracy. For example, the paper [6] combines the correlation coefficient, YUV histogram difference and running average difference to detect the shot boundary, although it can achieve a good result, the overhead is huge. Speeded up robust features (SURF) is an improvement on Scale-invariant feature transform (SIFT). It has several good characteristics, such as high speed and stable algorithm. It also can detect more f
Data Loading...
