Extracting a small number of key-frames that can abstract the content of video is very important for efficient browsing and retrieval in video databases. Most research on video content involves automatically detecting the boundaries between camera shots. After shot boundary detection there is a need for shots indexing. In this paper, we present the gradient field based algorithm of shot detection and new method of key-frames determination. We provide a novel algorithm aimed to find a compact set of key-frames that can represent a video segment for a given degree of fidelity. The advantages of proposed approach are high performance of detection of the key-frames and linear speed of retrieval from the databases. Also we provide a new concept of the shot segment analysis and interpretation, using graphical and numerical methods.