On the Development of Measurement and Emotion Information Processing Systems based on Chaosand Fractal Properties

在本研究中，我们开发了一个基于混沌分形分析的情绪测量系统。在实践中，情感分形维数分析方法是利用分形维数分析提取脑电图特征，从而最终评价人的情感，即感性。众所周知，这种方法可能无法获得比愤怒、喜悦、放松和悲伤等基本情绪更高级的情绪的足够识别。要实现21世纪与人类情感相关的产品生产，就需要对人类情感进行定量评价。因此，在本研究中，我们试图开发测量人类情绪的系统，并阐明相应的信息处理机制。此外，我们还研究了一个应用于人机界面的具有分形维数分析方法的EEG键盘等功能。在实验中，我们在屏幕上显示了6个目标，并检验了被试是否能从EEG的时空行为中识别出目标。结果表明，该方法的识别率可达~ 80%。因此，人们可以预期，本技术可能适用于脑电图键盘。此外，我们还成功地建立了一种基于近红外光谱（NIRS）来代替脑电图的人类情绪测量方法。因此，作为未来的一个问题，将这两种方法结合起来进一步提高识别率似乎是值得的。

In this research, we have developed an emotion measurement system based on chaos fractal analyses.In practice the emotion fractal dimension analysis method which may eventually evaluate the human emotion, i.e. Kansei, by means of extracting the features of electroencephalograph (EEG) with Fractal dimension Analysis. It has been well know that this method may not attain enough recognition of higher order emotion than the fundamental emotions such as anger, joy, relaxation and sadness. Evaluation of human emotion quantitatively is required to realize to produce manufactures in connection with the human emotion in the 21^<st> century. Therefore, in this research, we have attempted to develop the system to measure the human emotions as well as to clarify the corresponding information processing mechanism.In addition, we have examined an application to human interface to have a function such as EEG keyboard with Fractal-dimension Analysis Method. In practice we displayed six objects on the screen and examined whether a subject could recognize an intended object from the spatio-temporal behavior of EEG. As a result, it was found that the recognition rate was attained up to 〜80%. Therefore, one may expect that the present technology may be applicable to an EEG keyboard. Furthermore, we have succeeded to establish a method to measure the human emotion based on the near-infrared spectroscopy (NIRS) in stead of EEG. Therefore, as a future problem, it seems to be worthwhile to combine these two methods to improve the recognition rate further.