| Autism spectrum disorder(ASD),also known as autism,is a neurodevelopmental disorder that starts in infancy.Since scholars first proposed the disease in the last century,more and more people began to pay attention to the disease,but up to now,the core pathogenesis of the disease is still unclear,and there is still a lack of effective means to cure the disease.It is reported that the earlier the disease is found and treated,the greater the possibility of recovery and the less irreversible damage to patients.At present,the diagnosis of ASD mainly depends on professional doctors to evaluate the scale of patients,which is a complicated and time-consuming process.In recent years,more and more scholars began to use MRI and EEG to study the auxiliary diagnosis of ASD.However,these methods need to rely on expensive equipment and professional operators and specific environment,which limits the wide spread of these methods.This topic aims to solve this problem in the simplest and cheapest way.In order to study this topic,we collected 136 ASD children's videos and TD children's videos.For the videos we collected,we proposed two video classification algorithms based on eye movement track feature fusion for autistic children and multi feature fusion based on expression intensity and eye movement track for autistic children.The first algorithm mainly decomposes eye movement features into length features and angle features,and classifies them by feature fusion and other methods,which achieves better results than before,and the classification accuracy reaches 96.6%.Inspired by expression analysis,we propose an expression intensity estimation method to measure the expression intensity of ASD children and TD children.We also propose two methods,cumulative histogram and length direction histogram,to calculate the features.Experiments show that this method has a very good effect on the auxiliary diagnosis of autism.The proposed method of expression intensity is the first to be used in autism research. |
PDF Full Text Request