Real Environment Warning Model for Visually Impaired People in Trouble on the Blind Roads Based on Wavelet Scattering Network

When the visually impaired walk on the blind road, once they encounter obstacles to block the road, it will bring panic and even safety risks to the visually impaired. To solve this problem, based on the wavelet scattering network (WSN), this study proposes a real environment warning mode for visually impaired people when they are in trouble on the blind road. When the blind road is interrupted or obstructed, visually impaired people will experience tension or anxiety. In this study, Based on electroencephalogram (EEG) signals, this study uses the WSN method to identify the emotional state of visually impaired people, and then determine whether they need assistance. The wavelet scattering coefficients of EEG signals are extracted using the WSN method, resulting in the formation of a feature matrix. Subsequently, a support vector machine is employed for the purposes of classification and recognition. The results show that the recognition accuracy of this method reaches 95.11% in the three states of normal emotional state, general nervous state, and very nervous state of the created datasets. The classification accuracy on the SEED-IV dataset is 86.14%. The WSN method suggested in this study exhibits superior recognition accuracy and the quickest algorithm execution time when compared to previous emotion identification methods. In addition, compared with the no-warning model, the warning model proposed in this study can greatly reduce the time it takes for visually impaired people to wait for help. The WSN-based warning mode provides more reliable travel assistance for visually impaired people and reduces the risk of accidental injury.