Selection of auto-focus function in micro visual system

The algorithm of general focusing function is put forward, and a series of experiments is tested. Focusing character with different lighting and different content is validated and analyzed. The experiment result shows that the functions of gradient energy, Tenegrad, and Laplacian possess the stable, highly sensitive characteristic curve. It's better to use in small range with high accuracy; the functions of Variance and entropy have a large focusing range, good linearity and low sensitivity, so they are propitious to focus in large range with lower accuracy. So, there has better performance curve in background lighting than in reflection lighting. It's appropriate that using the functions of gradient energy or Tenegrad to focus for the image with more high frequency information, and using Laplacian function for the image with slightly high frequency information.