Facial expression recognition (FER) and its analysis becomes an attractive research study in the fields of computer vision applications and pattern recognition. These facial expressions are generally categorized into two kinds such as micro and macro-expressions. To detect the macro-expression effectively, an angle based pattern extraction models and Markov models are employed. But the micro- expression delivers more detailed information than the macro-expression. Other diffculties such as short durations and rapid spontaneous facial expression are induced due to the detection and analysis of the micro-expression. To solve these challenges, we propose the three novel techniques such as Active Shape Modeling (ACM), Random Walk (RW) and the Artificial Neural Network (ANN) which helps to improve the overall performance effectively. The key points from the facial expression over the video frames are predicted using ASM and are spatially associated with the original face through the procrustes analysis. Then the RW algorithm is used to learn the training features prior to ANN model. This RW is integrated with ANN model to improve the learning performance of micro-expression with minimum computation complexity. The experimental validation on two spontaneous micro-expression datasets such as Chinese Academy of Sciences Micro-Expression (CASME) and Spontaneous Micro-expression (SMIC) over the existing SVM classifiers shows its effectiveness in automatic micro– expression learning applications.

Facial expression recognition (FER) and its analysis becomes an attractive research study in the fields of computer vision applications and pattern recognition. These facial expressions are generally categorized into two kinds such as micro and macro-expressions. To detect the macro-expression effectively, an angle based pattern extraction models and Markov models are employed. But the micro- expression delivers more detailed information than the macro-expression. Other diffculties such as short durations and rapid spontaneous facial expression are induced due to the detection and analysis of the micro-expression. To solve these challenges, we propose the three novel techniques such as Active Shape Modeling (ACM), Random Walk (RW) and the Artificial Neural Network (ANN) which helps to improve the overall performance effectively. The key points from the facial expression over the video frames are predicted using ASM and are spatially associated with the original face through the procrustes analysis. Then the RW algorithm is used to learn the training features prior to ANN model. This RW is integrated with ANN model to improve the learning performance of micro-expression with minimum computation complexity. The experimental validation on two spontaneous micro-expression datasets such as Chinese Academy of Sciences Micro-Expression (CASME) and Spontaneous Micro-expression (SMIC) over the existing SVM classifiers shows its effectiveness in automatic micro– expression learning applications.