A New Sampling Method of Auto Focus for Voice Coil Motor in Camera Modules

摘要

We present an effective and efficient sampling table for auto focus searching algorithms to phone camera modules. Many searching algorithms base on focus value curve, and they do not consider the correlation between sampling and image characterization. Proposed table is calibrated by the camera images to meet results of optical lens, sensor response, noise, image pipeline, compressed method, and focus value function. We verify the algorithms in an image signal processor with mega-pixel level cameras. The table reduces total searching steps from 412 to 16, and images are still sharp. Our searching algorithm with the table can focus a scene within 0.6 seconds from infinity to 8 cm. For further progress, they can easily apply to mobile phones.