Video Temporal Segmentation Based on Color Histograms and Cross-Correlation

Several fields of knowledge generate and consume massive volumes of videos, such as entertainment, telemedicine, surveillance and security. The rapid growth in the demand for multimedia content has driven the development of fast and scalable mechanisms for storing, retrieving and transmitting video sequences. The automatic temporal segmentation is a fundamental process in the analysis of video content. This work proposes and evaluates an adaptive video shot detection based on color histograms and normalized cross-correlation. Experiments conducted on several video sequences demonstrate that the combination of these two features achieve high accuracy rates.