Study on performance evaluation and frequency characteristic for real-time image stabilization system

In image stabilization systems based on motion detection and real-time compensation, the time of image acquisition, and motion estimation, etc. resulted in delay of compensation signal in relation to actual vibration, and the influence gradually became more significant with the increase in vibration frequency. We presented a concept of compensation degree to quantitatively evaluate and analyze the influence and used Least Squares Support Vector Machine (LS-SVM) to predict the motion, so the signal delay was effectively correction. The experiment results showed that the LS-SVM had high prediction accuracy, fast calculation speed and good stability so it was in favor of improving the frequency characteristic of image stabilization systems.