Facial Expression Recognition Using Neural Networks and Log-Gabor Filters

Abstract

This study proposes a classification-based facial expression recognition method using a bank of multilayer perceptron neural networks. Six different facial expressions were considered. Firstly, logarithmic Gabor filters were applied to extract the features. Optimal subsets of features were then selected for each expression, down-sampled and further reduced in size via Principal Component Analysis (PCA). The arrays of eigenvectors were multiplied by the original log-Gabor features to form feature arrays concatenated into six data tensors, representing training sets for different emotions. Each tensor was then used to train one of the six parallel neural networks making each network most sensitive to a different emotion. The classification efficiency of the proposed method was tested on static images from the Cohn-Kanade database. The results were compared with the full set of log-Gabor features. The average percentage of the correct classifications varied across different expressions from 31% to 85% for the optimised sub-set of log-Gabor features and from 23% to 67% for the full set of features. The average correct classification rate was increased from 52% for the full set of the log-Gabor features, to 70% for the optimised sub-set of log-Gabor features.

Emotion Recognition System by a Neural Network Based Facial Expression Analysis

“…Based upon those results, it can be concluded that the accuracy of emotion recognition of the developed system is in line with those from other sources where a lot more samples were used in the creation and learning process, as in [9], [12] and [17]. Due to the difference between the databases used, a detailed comparison with those developed systems requires a more rigorous comparative analysis.…”

“…Generally, when considering emotion recognition from static images there are two basic principles that determine which facial region is going to be used for feature extraction. In one principle, the face is globally analyzed and features from the entire face are gathered as in [9] and [17], while according to the second principle, analysis is employed on only certain facial regions of which we know that they hold information about emotions, i.e. regions like the mouth, eyes or eyebrows.…”

Texture Fusion for Batik Motif Retrieval System

“…The singularity of the log function Log Gabor filters basically are defined in the frequency domain as Gaussian functions that shift from the origin [17]. Gabor filters present a limitation in the bandwidth where only bandwidth of 1 octave maximum could be designed [18][19][20]. Log Gabor consists of a logarithmic transformation in the Gabor domain which eliminates the DC-component allocated in medium and high-pass filters.…”

A simple automatic facial aging/rejuvenating synthesis method

“…The Gabor wavelet employs multi-channel and multi-layer methods to capture facial features while simultaneously effectively identifying skin surface topographies of different frequencies on human faces. It can also be used in classifying facial expressions [9] [10]. Gabor Wavelet is defined as:…”

“…The Gabor wavelet is mainly used in face recognition systems to capture frequency information from facial features [5][7] [9][10] [11]. The Gabor wavelet employs multi-channel and multi-layer methods to capture facial features while simultaneously effectively identifying skin surface topographies of different frequencies on human faces.…”