Automated TV Commercial Tracking and Auditing

Media stations, Auditing firms and commercial publishers are tended to audit the evidence of commercials, which had been published. This tracking system audits full screen television advertisements and radio advertisements. The research consists with three main components namely video boundary detection and feature extraction component, frame matching and searching component and audio matching module and validation. A novel method is proposed for abrupt video shot boundary detection comparing difference measures between adjacent frames and scoring features. Feature extraction and searching method was proposed to search advertisements in live stream using SURF descriptors and FLANN matching. Audio fingerprinting, Cross correlation and salient features are used to develop a novel audio matching mechanism with a domain specific dynamic matching score.