Rotation Invariant Multiple Face-detection Architecture for Smart TV

Abstract

This paper suggests a design of high quality real-time rotation face detection architecture for gesture recognition of smart TV. For high performance rotated face detection, the multiple-MCT(Modified Census Transform) architecture, which is robust against lighting change, was used. The Adaboost learning algorithm was used for creating optimized learning data. The proposed hardware structure was composed of Color Space Converter, Image Resizer, Noise Filter, Memory Controller Interface, Image Rotator, Image Scaler, MCT Generator, Candidate Detector, Confidence Switch, Confidence Mapper, Position Resizer, Data Grouper, Overlay Processor and Color Overlay Processer. As a result, suggested face detection device can conduct real-time processing at speed of at least 30 frames per second.