Bionic vision autofocus method based on a liquid lens

Digital imaging systems (DISs) have been widely used in industrial process control, field monitoring, and other domains, and the autofocusing capability of DISs is a key factor affecting the imaging quality and intelligence of the system. In view of the deficiencies of focusing accuracy and speed in current imaging systems, this paper proposes a fast autofocus method of bionic vision on the basis of the liquid lens. First, the sharpness recognition network and sharpness comparison network are designed based on the consideration of a human visual focusing mechanism. Then a sharpness evaluation function combined with the distance-aware algorithm and an adaptive focusing search algorithm are proposed. These lead to the construction of our proposed autofocus method with the introduction of the memory mechanism. In order to verify the effectiveness of the proposed method, an experimental platform based on a liquid lens is built to test its performance. Experiment confirms that the proposed autofocus method has obvious advantages in robustness, accuracy, and speed compared with traditional methods.