In this paper, the human emotions are analyzed from EEG Signal (Electroencephalogram) with different kinds of situation. Emotions are very important in different activity and decision making. Various feature extraction techniques like discrete wavelet transform, Higher Order crossings, Independent component analysis is used to extract the particular features. Those features are used to classify the emotions with different groups like arousal and valence level using different classification techniques like Neural networks, Support vector machine etc.. Based on these emotion groups analyze the performance and accuracy for the classification are determined.