An auto-focus method to determine an optimum position of a lens module includes defining search boundaries and an allowable difference, randomly sampling current values within the search boundaries d1, d2 and d3, and obtaining sharpness m1, m2 and m3. A parabola is determined where a vertex of the parabola is used as a next input current value where three larger values of the sharpness are used along with corresponding current values to re-determine a new parabola. Three current values are sampled and a corresponding current value is used to re-determine a new parabola and according to a comparison of the sharpness values, the lens module is driven to auto-focus.