Detection of position and posture of occupants using low resolution sensor

There has been an active development toward the extraction of human image by measuring heat sources with infrared rays. This is because infrared rays are more effective at detecting human than visible light. Detection of position and posture of occupants using low resolution sensor is proposed. The proposed system can measure the position and posture of occupants from data using a sensor. Other applications are: The surface temperature behavior of a steel specimen is done by obtaining the endurance state of the specimen by use of thermal images. This study presents a system for recognizing an armored vehicle by distinguishing marks and contours. This study presents an effective method for choosing features and processing images. A thermal image processing system consists of a thermal image sensor, an image pre-processor, and a neural network. The thermal image sensor is a pyroelectric type infrared image sensor, which does not require cooling and measures the thermal environment and extracts information of occupants. The improvement in the positional precision of the neural network by structuralization is examined. 1. One large neural network without structured.2. One posture-estimating neural network and two position-estimating neural networks where output is selected by the output of posture-estimating neural network 3. One posture-estimating neural networks and two position-estimating neural network where output is calculated from each two outputs (distance of sanding posture and sitting posture) and the outputs of the posture-estimating neural network.