A Visual Grammar Approach for TV Program Identification

TL;DR

This paper introduces a spatial-temporal visual grammar-based method for automatic TV program identification, constructing a catalogue of visual jingles and comparing stream signatures to recognize program types.

Contribution

It presents a novel approach using visual grammars for TV program identification, leveraging spatial-temporal analysis and similarity comparison to improve accuracy.

Findings

Effective identification of TV programs demonstrated on multiple streams.

High accuracy in matching visual jingles with program types.

Robustness across different channels and program variations.

Abstract

Automatic identification of TV programs within TV streams is an important task for archive exploitation. This paper proposes a new spatial-temporal approach to identify programs in TV streams in two main steps: First, a reference catalogue for video grammars of visual jingles is constructed. We exploit visual grammars characterizing instances of the same program type in order to identify the various program types in the TV stream. The role of video grammar is to represent the visual invariants for each visual jingle using a set of descriptors appropriate for each TV program. Secondly, programs in TV streams are identified by examining the similarity of the video signal to the visual grammars in the catalogue. The main idea of identification process consists in comparing the visual similarity of the video signal signature in TV stream to the catalogue elements. After presenting the proposed approach, the paper overviews the encouraging experimental results on several streams extracted from different channels and composed of several programs.