The work areas for emotion recognition are facial expressions, vocal, gesture and physiology signal. Facial expressions are one ofmost functional areas for face emotion recognition. For best results we should similar eye and lip as regular and irregular ellipse. The mainpurpose of this paper is introducing an Imperialist Competitive Algorithm (ICA) to optimize eye and lip ellipse characteristics. Thenperformance of three optimization methods including Imperialist Competitive Algorithm (ICA), Particle Swarm Optimization (PSO) and GeneticAlgorithm (GA) for this issue will be discussed. This process involves three stages pre-processing, feature extraction and classification. Firstly aseries of pre-processing tasks such as adjusting contrast, filtering, skin color segmentation and edge detection are done. One of important tasks atthis stage after pre-processing is feature extraction. Projection profile method to reason has high speed and high precision used in featureextraction. Secondly ICA, GA and PSO are used to optimize eye and lip ellipse characteristics. Finally in the third stage with using featuresobtained on optimal ellipse eye and lip, emotion a person according to experimental results have been classified. The obtained results show thatsuccess rate and running speed in ICA is better than PSO and these two parameters for PSO are better than GA.Keywords: Face emotion recognition, Projection profile, Imperialist Competitive Algorithm (ICA), Particle Swarm Optimization (PSO)algorithm and Genetic algorithm (GA).
Performance Optimization Algorithms in Classification Face Emotion Recognition
13.01.2017
doi:10.26483/ijarcs.v3i1.1022
International Journal of Advanced Research in Computer Science; Vol 3, No 1 (2012): January-February 2012; 127-130 ; 0976-5697 ; 10.26483/ijarcs.v3i1
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Europäisches Patentamt | 2024
|Vehicle-mounted human-computer interface with face and emotion recognition function
Europäisches Patentamt | 2020
|