Description

The sign is recognize is useful to the handicapped peoples to speech with others in world. The sign recognize is difficult to process because the hand shape identification is difficult. In recent year there are many research will goes to recognize the sign easily. Recognition of the letters or sentences from signs has gained importance in the education system of deaf and dumb people. The main objective of the process is the identification of the exact Arabic letters or words from the sign image that is given as input. To extract the exact sign region from the input image skin pixel identification based on thresholding is proposed. To extract the visual features from the input image the mean and standard deviation values were used. To recognize the words using modified KNN classifier and measure the performance of the classifier. A method that identifies the Arabic letters and sentences based on sign language is proposed. The process uses the images that depicting the sign of the Arabic word or letters taken in a clear background is the input for the proposed process. The images were converted to gray and the sign regions alone were extracted from the image based on the thresholding process. The threshold is identified based on the maximum pixel. The visual features were extracted from the image and the extracted features were then used for the recognition process. The recognition process is done based on modified KNN algorithm. The performance of the process is measured. The performance of the process indicates that the proposed approach is more improved compared to the existing methods for the recognition process. The complexity is much reduced since the process does not include the glove based reading and other hardwares. The process is simple and reliable due to the application of the simple visual based features from the input image. The recognition rate of the proposed approach indicates the number of the words which are exactly identified.