Digital Image Stabilization Using a Functional Neural Fuzzy Network

This study proposes a real-time video stabilization method to eliminate unwanted vibration, preserve the intended movement of camera, and improve the stability of the captured video sequence. The proposed method uses a functional neural fuzzy network to learn the characteristics of different vibrations and then choose the adequate compensation weight for two different methods to calculate the correction vector. Experimental results show that the proposed method has superior performance over other motion compensation methods.