Emotion detection is one of the major research areas in human-machine interface design. With emotions playing an important role in today’s technology revolution, machines are also designed to understand and respond to human emotions. Next generation computers and multimedia gadgets and gaming gadgets are designed to identify human emotions and accordingly perform the activities. Emotions in humans are classified into valence and arousal behaviours. Stress is one of the components that influence emotions. True emotions exhibited by humans are dominated by stress levels. Determining emotions also can provide indirect measurement of stress levels. The focus of this research work is to detect emotions and classify emotions into normal and abnormal emotions indicating true emotions and stress related emotions. EEG signals that are recorded from brain sensors are processed and decomposed into wavelet domain and energy levels are determined. The wavelet energy levels are processed by the neural networks structure to classify the wavelet energy levels into normal and abnormal levels indicating the emotions. MATLAB environment is used for modelling of proposed emotion detection and classification algorithm considering EEG signals recorded from more than 40 subjects. The emotion detection and classification algorithm is designed for implementation on FPGA platform optimizing area, timing and power performances.