An effective hybrid ABC-CS optimized ANN classifier for facial expression recognition

The necessity of identifying an individual's intentions solely through facial expressions, without verbal communication, is crucial in daily life. Various approaches employing Artificial Intelligence interventions are utilized to recognize emotions. Consequently, the present study aims to propose an optimal image processing approach named as Hybrid Optimization Based Artificial Neural Network (HOBANN) for the recognition of facial expression. In this process, the implementation of Adaptive Gabor filter is preferred to remove the noise contents in the input image while preserving edge information effectively. Following this, Cascaded Fuzzy C-Means (CFCM) approach is employed to segment the image into multiple sections, enhancing accuracy across an extensive range. In the process of feature extraction, an efficient method known as Gray-Level Co-occurrence Matrix (GLCM) is employed. This method offers several benefits, including reduced complexity and optimal computations of pixel variance. Additionally, a novel hybrid optimization technique called Artificial Bee Colony-Cuckoo Search (ABC-CS) is introduced. This hybrid optimization is applied in the final classification process using an Artificial Neural Network (ANN) classifier, aiming to recognize expressions with maximum accuracy effectively. The Matlab simulink is preferred to validate the entire study and the simulated outcomes have demonstrated that the overall performance of this approach is significantly high as it delivers optimal accuracy rate of 97.93%.