Road Bump Outlier Detection of Moving Videos Based on Domestic Kylin Operating System

Abstract

With the increasing number of moving videos, anomaly detection of moving videos has become a popular data mining task in the field of intelligent transportation. Traditional road anomaly detection algorithms are hard to detect road bump outliers while the domestic platform has not yet applied road bump detection methods using the accelerometer and gyroscope data. For this, we proposed a road bump outlier detection algorithm (RBOD) and illustrated migration and the improvement of our algorithm for the domestic platforms. Our RBOD algorithm used a Kalman Filter-based method to solve the noise data problem of the accelerometer or gyroscope and selected the accelerometer or gyroscope data for outlier detection according to the sampling frequency. The experimental results show that our RBOD algorithm can detect moving things anomalies efficiently and accurately.