Research On The Evaluation Method Of Working Memory Load Based On Facial Video

With the rapid development of human society,the working memory load generated in people’s daily life will lead to people’s physiological and psychological changes.Therefore,the evaluation of working memory load is of great significance.In the existing working memory load evaluation methods,the subjective evaluation method is simple and convenient to evaluate through the subjective feelings of people,but it has a high degree of subjectivity;the objective evaluation method is mainly evaluated by observing the changes of physiological characteristics.This evaluation method has high accuracy,but most of them need to use contact equipment and have a poor sense of experience.Imaging Photoplethysmography(IPPG)technology,which has emerged in recent years,can extract physiological features in a non-contact way,but how to extract high-quality IPPG signals has become a research difficulty.To solve the above problems,this paper proposes a working memory load evaluation method based on facial video by using IPPG technology.The main research work and contributions are as follows:(1)For the high-quality extraction of IPPG signals,this paper proposes a joint denoising algorithm based on TVFEMD and wavelet threshold.The algorithm first uses the TVFEMD algorithm to decompose the IPPG signal with more noise,and obtains multiple Intrinsic Mode Function(IMF)components.The processed IMF components are used for signal reconstruction.Finally,after filtering,the IPPG signal with higher quality is extracted.(2)For the extracted IPPG signal,calculate the heart rate closely related to the working memory load,and carry out error analysis,which proves the validity and accuracy of the noncontact heart rate measurement using the algorithm in this paper;in addition,this paper analyzes the influence of different frame rates and different spatial pixel sampling numbers on the noncontact heart rate measurement,which has certain exploration value for the follow-up IPPG technology research.(3)Aiming at the problems of current working memory load evaluation methods,this paper proposes a non-contact objective evaluation method of working memory load based on facial video.The method uses facial video to extract IPPG signals,and obtains features such as heart rate variability.Considering the influence of feature redundancy,recursive feature elimination algorithm is used for feature selection,and the obtained feature subset is input into the random forest to classify the working memory load level.A high classification accuracy is achieved,thereby establishing the mapping relationship between various features and the level of working memory load,and realizing an objective evaluation of working memory load.