A new approach incorporating adaptive lighting intensity for micro-crack inspection of solar wafers with variable thickness is proposed. Wafer thickness is measured with a pair of laser displacement sensors and the lighting intensity is adaptively adjusted to normalize near infrared (NIR) transmission based on measured thickness. This technique enables the image contrast be maintained at relatively uniform intensity in response to the variation of the solar wafer thickness. An improved version of Niblack segmentation algorithm is developed for this application. Experimental results demonstrate the competitiveness of the proposed system compared with established techniques, and achieves better performance both visually and quantitatively. Meanwhile, the runtime is consistently less than 1 s, corresponding to a throughput rate of approximately 3600 wafers/h. These results suggest that the methods and procedures are suitable for online processing of solar wafers.