Sign Language Recognition System Sign Wave

Abstract – People with hearing and speech disabilities often struggle to communicate without a translator, as sign language is not universally understood. They rely heavily on hand gestures for non-verbal communication. To address this, the paper proposes a system for automatic recognition of finger spelling in Indian Sign Language. The process begins by capturing the sign as an image input. Skin colour-based segmentation is used to detect hand shape, followed by conversion to a binary image. A Euclidean distance transformation is applied, and row/column projections are performed. For feature extraction, central moments and HU’s moments are used, while classification is done using Neural Networks and SVM. Keywords: Artificial Neural Network, Central Moments, Distance Transformation, Fourier Descriptor, HU’s Moments, Indian Sign Language, Projection, Skin Segmentation, SVM.