JELE_V1_N4_RP7 Goh Chia Chieh Dino Isa Journal on Electronics Engineering 2249 – 0760 1 4 43 49 Real-time Emotion Recognition, Parallel Computing, CUDA, Thermal Imaging, Heart Rate Monitoring Real-time human emotion recognition is a challenging task. Speed and accuracy has always been the main concern for human emotion analysis. Such system poses an even greater challenge for implementation in a limited space of a constantly mobile environment, for example, in a moving vehicle. On top of that, power consumption is another major issue to be considered. This paper proposes an efficient method on how such system can be implemented in the mobile environment by utilizing parallel CPU and GPU to achieve real-time emotion analysis and judgment. Using low power consumption Single Board Computer (SBC) with a dual graphic card installed, the task will be split into multiple threads which process visual, thermal and heart rate emotion monitoring system simultaneously. Thus, better accuracy and optimal performance is attained through parallel GPU and CPU processing. Handling enormous of data through parallel processing requires expertise in combining the CPU and GPU tasks together by merging different kinds of data competently. June - August 2011 Copyright © 2011 i-manager publications. All rights reserved. i-manager Publications http://www.imanagerpublications.com/Article.aspx?ArticleId=1510