Fallen-down real-time detection method and timing gmm Model

A real-time detection method and timing model falls GMM based, comprises the following steps: 1) background modeling; 2) morphological processing; 3) moving target detection; 4) determines the aspect ratio change body; 5) configured to change the focus state matrix; 6) exponential smoothing update transactions; 7) falls to a predetermined threshold matrix; 8) according to the test data, determined in accordance with the formula a fall detection. The present invention employs the focus state matrix characterizing information on the body motion is decomposed into a plurality of parameters, and give a higher weight perpendicular variable weight, the body state change detection more accurately; using weights based on exponential smoothing time series of historical data weight adjust the current cumulative data more reasonable. The present invention is applicable to both the ground is also applicable to stairs and other environmental ramp environment.