An improved CRITIC-LRA method for EEG cognitive image evaluation

Authors

  • Jiqiang Qu, Guoqing Wang School of Mechanical Engineering, Qilu University of Technology (Shandong Academy of Sciences), Jinan 250300, China

Keywords:

Event-Related Potentials, image evaluation, CRITIC-LRA

Abstract

In the process of image evaluation, the opposition and entanglement between different modeling elements in the target scheme make the product show composite image characteristics. Reviewers are easily affected by subjective factors, and it is difficult to measure implicit factors such as emotion and cognition, which leads to the distortion of evaluation results. Event-Related Potentials (ERP) can accurately obtain individual image perception processes and provide objective physiological data. This study combines ERP with the improved CRITIC-LRA (Criteria Importance Though Intercriteria Correlation - Linear Regression Analysis), implements the evaluation of automobile shape image, introduces the Spearman correlation coefficient to calculate the conflict between indicators, uses LRA to reduce the correlation between indicators, and solves the problems of requiring normal distribution of data and ignoring the internal correlation of indicators in the CRITIC, making it suitable for small sample size Evaluation of nonlinear data. The results show that the improved CRITIC-LRA method can improve the differentiation of image indicators, get the implicit image weight value, and provide quantitative, objective, and valuable evaluation results for reviewers

Downloads

Published

2023-06-28