Magnifying emotions recognition from facial expression is highly demanded in several applications domains such as security, education, psychology, medical diagnosis, marketing and business negotiations. For the growth and productivity of these domains, researchers are keenly involved in improving the effectiveness of facial expression recognition (FER) systems. However, they still lack potency in terms of recognition accuracy, inter-subject facial variations and appearance complexity. This paper attempts to improve recognition accuracy by employing Radon transform and Gabor wavelet transform along with robust classifiers. Facial detection is examined by oval parameters approach and facial tracking is achieved using vertex mask generation. Radon transform and Gabor transform filters have been applied to extract variable features. Finally, self-organized maps and neural network are used as recognizer engine to measure six basic facial expressions. Unlike conventional results that were evaluated using a single dataset, our experimental results have shown state-of-the-art accuracy of 86 and 83.7 percent over two public datasets as Cohn-Kanade and AT&T datasets respectively.