Neoteric Video Segmentation Scheme using NCD for Detecting Indistinct Scenes

In this paper, an implementation of a neoteric and efficient video segmentation scheme aimed to detect interlaced video scenes using the normalized compression distance (NCD) is proposed. NCD approximates the value of a theoretical distance between two adjacent video frames using the Kolmogorov complexity theory. This proposed scheme presents a robust and efficient automatic video segmentation based on pervious researches which lack the accuracy when dealing with cuts, fades, or morphing as in sports, commercials and news clips where fading can be greatly noticed. The experimental results gave back a stunning recall and a high precision values. This scheme indicates an overall improvement on video segmentation results achieved in reference to our previous approaches.