A method and apparatus of capturing non-structured light images and structured light images for deriving depth information are disclosed. According to the method, one or more non-SL (non-structured light) images without structured light and one or more initial SL (structured light) images formed on a common image plane are captured by projecting structured light patterns in a visible spectrum with the structured light source adjusted to generate initial structured light at an initial intensity level. The signal quality of structured light patterns reflected from one or more objects is evaluated based on the non-SL images and the initial SL images. If the signal quality of structured light patterns is below a threshold, a next set of SL images are captured by increasing the structured light level from a previous level until the signal of the structured light patterns is satisfactory.