CHILD AUTISM DETECTION BASED ON FACIAL FEATURE CLASSIFICATION

Autism Spectrum disorder is a congenital anomaly that results in recurrent behavior pattern in children and conjointly spoils their social interaction. It is proven that there is strong association between autism and congenital anomalies. Autistic people have issues in following routines and they are unable to predict the outcomes of certain behaviors. Also, in challenging environments the autistic persons express anger, sadness or frustration. A large body of evidence suggests that several method sare available for diagnosis of autism. But they are costly or time consuming. In this article we suggest a new method based of abnormal facial structural clues for quick autism diagnosis. This work analyses the if facial features such as an eye, nose and lip distance in a child image and its arrangements can be an autism indicator. These locations classified using the classifier such as the SVM, KNN and the Naive Bayes classifier. Also, we discuss few algorithms and steps to calculate the contour extraction such as area and perimeter with clip limited adaptive algorithm. The feature extraction involves the texture analysis of the face. This work is simulated using the MATLAB R 2014b software and the results has been analyzed using the accuracy, sensitivity and specificity of the classifier