Real-Time Human Tracker Based on Location and Motion Recognition of User for Smart Home

The ubiquitous smart home provides an automatic home service via analysis of human and home contexts. It receives a lot of contexts from the human and the home environment, but the most important context is information about the human’s location and motion. In this paper, we present a real-time human tracker that detects human location and motion using cameras and predicts services in the ubiquitous smart home. We propose an algorithm that captures effective area from four network cameras in which human is detected and his position is estimated. To detect human motion, three kinds of images are used: IMAGE1: empty room image, IMAGE2: image of furniture and home appliances in the home, IMAGE3: image of IMAGE2 and the human. The system decides if specific furniture or home appliance is associated with the human by analyzing three images, and estimates human motion using a support vector machine (SVM). Human motion was recognized as four types: "lie down", "sit", "stand-up", and "walk". The human motion recognition is decided from the pixel number by the array line of the moving object using SVM. We evaluated each motion 1000 times. The average accuracy of all the motions was found to be 85.3%.