Scene break detection: a comparison

Abstract

The automatic organization of video databases according to the semantic content of data is a key aspect for e cient indexing and fast retrieval of audiovisual material. In order to generate indices that can be used to access a video database, a description of each video s e quence i s n e c essary. The identi cation of objects present in a frame and the track of their motion and interaction in space and time, is attractive but not yet very robust. For this reason, since the early 90's, attempts have been applied in trying to segment a video in shots. For each shot a representative frame of the shot, called k-frame, is usually chosen and the video can be analyzed through its k-frames. Even if abrupt scene changes are relatively easy to detected, it is more di cult to identify special effects, such as dissolve, that were o p erated in the editing stage to merge two shots. Unfortunately, these special e ects are normally used to stress the importance of the scene change from a content point of view, so they are extremely relevant therefore they should not be missed. Beside, it is very important to determine precisely the beginning and the end of the transition in the case of dissolves and fades. In this work, two new parameters are proposed. These characterize the precision of boundaries of special e ects when the scene change involves more than two frames. They are combined with the common recall and precision parameters. Three types of algorithms for cut detection are c onsidered: histogram-based, motionbased and contour-based. These algorithms are tested and compared on several video s e quences. Results will show that the best performance is achieved by the global histogram-based method which uses color information.

Detection and Classification of Shot Transitions

“…The histogram-based method was chosen since it is a well established technique and has been shown to perform well in detecting edit effects [5]. The feature-based method was chosen for its ability to detect gradual transitions, although its performance was reported on limited test sequences [9].…”

“…The histogram-based method we used is similar to the method with the best performance in the comparative investigation by Lupatini et al [5]. This approach uses the 2 value to define the difference between two global colour histograms which is compared against two thresholds, T H and T L .…”

“…A frame pair where this difference is greater than some predefined threshold is considered to contain a shot cut. Histogram-based methods are the most common approach to shot cut detection in use today, since they offer a good trade-off between accuracy and computational efficiency [1,5,6,10]. Other approaches can be grouped into pixel-based methods [10] and region-based methods [6,10].…”

“…Most of these are weakened by large object and camera motions. Hence, motionbased algorithms have been proposed to distinguish between differences caused by motion and those caused by an abrupt transition [5,1,7].…”

“…In the case of shot cuts, the content change is usually large and easier to detect [4,5]. However, the inter-frame difference during a gradual transition is usually small.…”