Abrupt Shot Changes Research Articles

Two Stage Shot Boundary Detection via Feature Fusion and Spatial-Temporal Convolutional Neural Networks

Shot boundary detection is essentially to detect the position of frames where the shot changes. It has been actively studied in video analysis and management for convenience, which becomes a key technique with the rapid proliferation of rich and diverse videos. With respect to the complex characteristics of different shots in varying length and content variation property, in this paper we present a two stage method for shot boundary detection (TSSBD) which distinguishes abrupt shot by fusing color histogram and deep features, and locate gradual shot changes with C3D-based deep analysis. Abrupt shot changes are detected firstly as it occurs between two frames, which divides the complete video into segments containing gradual transitions; Over these video segments, gradual shot change detection is implemented using 3D-convolutional neural network, which classifies clips into specific gradual shot change types; Finally, an effective merging strategy is proposed to locate positions of gradual shot transitions. The experimental analysis illustrates that the proposed progressive method is capable of detecting both abrupt shot transitions and gradual shot transitions accurately.

A Supervised Learning Approach to Flashlight Detection

ABSTRACTShot boundary detection is a fundamental step of video indexing. One crucial issue of this step is the discrimination of abrupt shot change from flashlight, because flashlight often induces a false shot boundary. Support vector machine (SVM) is a supervised learning technique for data classification. In this paper, we propose a SVM-based technique to detect flashlights in video. Our approach to flashlight detection is based on the facts that the duration of flashlight is short and the video contents before and after a flashlight should be similar. Therefore, we design a sliding window in temporal domain to monitor the instantaneous video variation and extract color and edge features to compare the visual contents between two video segments. Then, a SVM is employed to classify the luminance variation into flashlight or shot cut. Experimental results indicate that the proposed approach is effective and outperforms some existing techniques.

Enhanced shot boundary detection using video text information

Shot boundary detection is a pre-requisite process for content-based video indexing and retrieval, an application domain that has attracted much research and consumer attentions due to the explosive growth of video data available and accessible to users worldwide. Currently, a number of edge-based techniques have been proposed in the literature for detecting abrupt shot boundaries to avoid the influence of flashlights common in many video types, such as sports, news, entertainment, and interviews videos. However, these techniques are susceptible to miss and false detections of abrupt shot changes. Our study shows that one main reason for these errors is due to the presence of superimposed texts that are common in various video genres. To address this problem, we present in this paper an efficient method that utilizes the edge type - text-edge (edge in text region) or non-text-edge (edge in non-text region) - to reduce erroneous detection due to the sudden appearance and disappearance of superimposed text. Extensive experiments have been conducted and the results show that our proposed method, as compared with other existing methods, can obtain better performance in detecting abrupt shot boundaries and differentiating them from the effects of superimposed text.

Automatic Shot Change Detection Algorithm Using Multi-stage Clustering for MPEG-Compressed Videos

Automatic shot change detection has been recognized as an important research issue for video classification. This paper proposes an automatic clustering-based algorithm for shot change detection in MPEG-compressed videos with a small number of user-defined parameters. For accurate detection of abrupt and gradual shot changes, the proper selection and extraction of features are important. We first propose a fast edge image extraction scheme in the DCT domain on the basis of AC prediction. Then, by using the features extracted from the edge images and DC images, a two-stage clustering-based algorithm is proposed for shot change detection. In the first stage, the algorithm detects abrupt shot changes by employing two-means clustering on the 2-D feature space of histogram and pixel differences between two neighboring DC frames. In the next stage, it subsequently explores gradual shot changes between two adjacent abrupt shot changes by performing a two-step clustering scheme, which uses multiple features such as an edge energy diagram and several frame difference measures. Simulation results show that the proposed algorithm is fast and accurate.

Shot change detection from partially decoded MPEG data

In this paper, we propose a scene decomposition algorithm from MPEG compressed video data. As preprocessing for scene decomposition, partial reconstruction methods of DC images for P- and B-pictures as well as I-pictures directly from an MPEG bitstream are used. As for detection algorithms, we have exploited several methods for detection of abrupt shot changes, dissolves, and wipe transitions using comparison of DC images between frames and coding information such as motion vectors. A method is also proposed for exclusion of undesired detection such as flashlight scenes in order to enhance shot change detection accuracy. More than 95% decomposition accuracy has been obtained in an experiment using a TV program longer than 1 hour. It is also found that in the proposed algorithm, shot change detection can be performed more than 5 times faster than normal playback speed using a 130MIPS workstation. © 1999 Scripta Technica, Syst Comp Jpn, 30(8): 11–22, 1999