Feature Extraction And Classification For Light Sleep Stages

Sleep is necessary for human physical activity, and sleep can make people recover from the fatigue state and remain in good condition. All night sleep is one of the main sleep models, which is the main way to restore health, but short time sleep during the day also plays an important role to restore health, maintain good mental state, and improve the quality of rest.In this paper, the study subject is short sleep during the day, which is mainly to solve the stages issues of short time sleep during the day. During the day short time sleep, the environment which the subjects in are different from the night, for example, the light during the day is strong, but is weak in the night; the noise outside during the day play a greater impact on the subjects, and there are almost no noise at night; the temperature in the daytime is high, which is low at night; besides the difference between the subjects is also obvious, some subjects have stable sleep state, but others may have significantly volatile sleep state. According to the characteristics of short time sleep during the day, the paper is mainly to study feature extraction method and classification method for light sleep stages.Firstly, the paper analyzes and extracts the parameters of various stages of light sleep. The paper analyzes the data of the short time sleep during the day, including EEG (Electroencephalography, EEG), EOG (Electrooculography, EOG), and EMG (Electromyography, EMG). According to the differences of sleep beteewn the day and night, and the difference of EMG is not obvious in various stages of the short time sleep, eventually, we eliminate redundant parameters, select the appropriate parameters EEG and EOG, and use these two parameters as the characteristics signals for short term sleep during the day. According to the result, these parameters can better reflect the trends in changes of sleep depth during short time sleepSecondly, based on the extracted features, combined with SVM, we design and implementation the stages classification method for light sleep that can accurately distinguish the sleep stages appeared in short time sleep during the day. According to these formulas for calculating the parameters constructed in the paper, we use EEG and EOG as a classification parameter, then select the training data and test data, and apply SVM for the classification of short time sleep. SVM is valid for the study of short time sleep during the day according to the result.