The Importance Of Sign Language Recognition

772 Words4 Pages
ABSTRACT Sign Language Recognition plays an important role in Human Computer Interaction system. This research focuses to implement sign language recognition system for Tamil language. In previous works, captured images are in a high intensity environment which is held at black background so there is a need to avoid shadow effects. But this system proposes real time recognition of video image with existing background, to identify 32 Tamil alphabets which include vowels and consonants. The palm image is captured, preprocessed by applying active contours for background subtraction, Feature is extracted based on the position of five fingers (up/down) , the distance is calculated between centre point of palm and point in finger tip, If the distance…show more content…
Sign Language Recognition is one of the multidisciplinary research area which involves computer vision, pattern recognition, image segmentation and natural language processing. This system requires the knowledge of hand shape, hand motion, position and orientation. Several regional languages exist around the world American Sign Language (ASL), French Sign Language (FSL), German Sign Language (GSL), Indian Sign Language (ISL) etc[9]. This research focuses on one of the South Indian Language, Tamil. One of the challenging task is to track the hands to extract features. In [1] and [2], the signer is required to wear colored gloves to simplify the tracking and segmentation of the hands. But this is an unnatural way to sign. A more natural approach uses skin color for segmentation of the hands. The proposed approach is designed to recognize 32 alphabets in Tamil which includes vowels and consonants. II.METHODOLOGY The proposed system is designed with five processing steps that are Image acquisition, Preprocessing, Feature extraction, Training and Testing. The architecture is shown in following…show more content…
Background subtraction technique [3] detects the movement of regions in an image by calculating the difference between current image and the reference background image in a pixel-by-pixel fashion. It segregates foreground image from the background image. Binarization is applied to convert RGB color image to black and white image. Then active contour is used to extract the outline of palm image which are in turn converted to RGB image[10]. FEATURE EXTRACTION The centre point of the palm is taken as a reference point. The image is scanned to identify the finger tip position. The distance between the reference point and finger tip of five fingers namely Little finger, Ring finger, Middle finger, Index finger and Thumb finger are calculated. Here threshold can be set that can be compared with calculated distance to identify the type of finger and its position either up or down. Binary value 1 is assigned if finger is in up position else 0 for down. TRAINING PHASE The captured image, binary value of image and its corresponding Tamil alphabet is stored in database. The database consists of 31 alphabets includes 12 vowels, 18 consonants and 1 ayutha eluthu with its binary value. To indicate

More about The Importance Of Sign Language Recognition

Open Document