The Fusion of Feature Extraction Applications and Blurring Techniques for Classifying Irish Sign Language

Sign language plays a crucial role in the non-verbal communication of a person with disabilities such as deafness and dumbness. The role of hand gestures for disabled people and its common use is undeniable. Therefore, image-based hand gestures were the most potential method to recognize sign language. This research attempted to combine feature extraction applications and blurring techniques for recognition. Not only the level of accuracy but also processing time was studied and analyzed when different feature-based extraction approaches. These methods include Principal Component Analysis (PCA) and Discrete Wavelet Transformation (DWT). Following this, the comparison between the three classification models named Support Vector Machine (SVM), K-Nearest Neighbor (k-NN), and Ensemble method was performed in this article. As a result, both feature extraction approaches with a particular blurring technique were reported with a performance of about 99%. In addition, the effectiveness of different blurring and image-cropping techniques was surveyed. These results were the first selection to find the most optimal methods for future work.