Face and Gesture Recognition for Human-Robot Interaction
Hasanuzzaman Hasanuzzaman, Haruki Ueno · 2007
Recognition 150dimensions: number of cameras, speed and latency (real-time or not), structural environment (restriction on lighting conditions and background), primary features (color, edge, regions, moments, etc.), etc.Multiple cameras can be used to overcome occlusion problems for image acquisition but this adds correspondence and integration problems.The aim of this chapter is to present a vision-based face and hand gesture recognition method.The scope of this chapter is versatile.Segmentation of face and hand regions from the cluttered background, generation of eigenvectors and feature vectors in training phase, classification of face and hand poses, recognizes the user and gesture.In this chapter we present a method for recognizing face and gestures in real-time combining skin-color based segmentation and subspace-based patterns matching techniques.In this method three larger skin like regions are segmented from the input images using skin color information from YIQ color space, assuming face and two hands may present in the images at the same time.Segmented blocks are filtered and normalized to remove noises and to form fixed size images as training images.Subspace method is used for classifying hand poses and face from three skin-like regions.If the combination of three skin-like regions at a particular frame matches with the predefined gesture then corresponding gesture command is generated.Gesture commands are being sent to robots through TCP-IP network and their actions are being accomplished according to user's predefined action for that gesture.In this chapter we have also addressed multi directional face recognition system using subspace method.We have prepared training images in different illuminations to adapt our system with illumination variation.This chapter is organized as follows.Section 2 focuses on the related research regarding person identification and gesture recognition.In section 3 we briefly describe skin like regions segmentation, filtering and normalization techniques.Section 4 describes subspace method for face and hand poses classification.Section 5 presents person identification and gesture recognition method.Section 6 focuses on human-robot interaction scenarios.Section 7 concludes this chapter and focuses on future research. How to referenceIn order to correctly reference this scholarly work, feel free to