Neonatal Pain Expression Recognition Based On Collaborative Representation

Due to the newborns can not complained of feeling pain, so a review of neonatal pain rely mainly on those medical staff who accepted the professional training. This creates a lot of time and energy consumption, and the results couldn’t make corresponding measures in time to the neonatal pain with the influence of subjective factors. So it’s necessary and meaning to establish a system which can automatically identify the expression of neonatal facial pain.For pain expression recognition of newborns, this thesis presents neonatal pain expression recognition based on collaborative representation. The main research work and achievements are as follows:(1) Research on the influence of different block sizes of recognition performance when using uniform pattern and symmetric pattern LBP feature extraction. Chunking recognition rate is better than not partitioned.(2) Because of the contribution of different facial expressions organs or regions for the classification is different, in order to highlight its expression characteristics, giving greater weight to higher level dependence areas.(3) Research on dimension reduction methods, such as PCA, 2D-PCA, 2D-LDA and so on. Analyze the relationship between each feature dimension and recognition rate.(4) Research on the classification method based on collaborative representation of neonatal pain expression.The experiments on neonatal expression image library show that when the training sample is 600, feature extraction using block based weighted LBP, classification methods using collaborative representation, obtain the highest average recognition rate of 87.50%. It proves that the collaboration representation classification method is effective and reliable in neonatal pain expression recognition.