Research and Development of Elderly People Surveillance Technology Based on Acoustic Situation Recognition

为保障独居老人的安全生活，研究了基于声学传感器的监控技术。这种方法是将日常生活中的声音事件转换为原始符号，并通过对符号的时间序列模式进行数据挖掘来识别情况和生活状况。在本研究中，我们开发了以下技术：多麦克风阵列；脚步识别；人的跟踪；声学新颖性检测；声学事件分类；声学特征提取；长期生活日志数据中的行为状况识别。我们还收集了样板房中老年人的生活行为数据进行评估。

To support the safe life of the elderly living alone, surveillance technology based on acoustic sensors was researched. The approach is converting acoustic events in everyday life into original symbols, and recognizing situations and living conditions by data mining of time series patterns of the symbols. In this research, we developed the following technologies: multiple microphone arrays; footsteps recognition; people tracking; acoustic novelty detection; acoustic event categorization; acoustic feature extraction; behavioral situation recognition from long term life-log data. We also collected life behavioral data of the elderly in a model house for evaluation.