Hand Gesture Recognition with Symmetric Pattern under Diverse Illuminated Conditions Using Artificial Neural Network

This paper investigated the effects of variant lighting conditions on recognition process. A framework is proposed to improve performance gesture under illumination using luminosity method. To prove concept, a workable testbed has been developed in laboratory by Microsoft Kinect sensor capture depth images for purpose acquiring diverse resolution data. For this, case study was formulated achieve an improved accuracy rate illuminated conditions. data preparation, American Sign Language (ASL) used create dataset all twenty-six signs, evaluated real-time The method uses set symmetric patterns as feature order identify human hands and recognize gestures extracted through hand perimeter feature-extraction methods. Scale-Invariant Feature Transform (SIFT) identification significant key points ASL-based with their relevant features. Finally, Artificial Neural Network (ANN) trained different environments classify utilizing selected features validation. experimental results showed that system performed well multiple pixel sizes. total aggregate 97.3% achieved across 26 alphabet datasets only 2.7% error rate, which shows overall efficiency ANN architecture terms processing time.