Novel Pose-Variant Face Detection Method for Human-Robot Interaction Application.

Abstract

We propose an e fective facial features detection method forhuman–robotinteraction (HRI) with indoor mobile robot.In case ofhuman-robotinteraction atmobile robot,its vision system has to cope with di ficult problemssuch aspose variations,illumination changes, and complexbackgrounds,which problemsmainly arise from the movementofmobile robot.In thispaper,in order to overcome such problems,we suggesta new facial feature detection approach based on local image region and directpixel-intensity distributions.Forthis,we propose two novel concepts;the directional template forevaluating intensity distributions and the edge-like blob map image with multiple strength intensity.Using this blob map image,we show thatthe locationsofmajorfacial