As quality control of micro devices and early detection of micro defects in these devices are becoming increasingly important, the demand for a fast and automated online inspection technique to detect micro defects with high spatial resolution is increasing. In this study, a reference-free micro defect visualization algorithm is developed based on laser scanning thermography to detect micro defects in devices instantaneously and automatically with high spatial resolution. A pulse modulated continuous wave laser generates thermal waves in a target device, and the corresponding thermal responses are recorded by an infrared (IR) camera. When the thermal wave encounters a micro defect, the propagation of the thermal wave is blocked at the interface of the micro defect. The blockage of the thermal wave is detected by the proposed reference-free micro defect visualization algorithm. First, an edge detection algorithm is applied to a raw thermal image obtained at a specific time point to extract the thermal discontinuities formed at the boundaries of the defect. The edge images obtained from all time sequences are then assembled into a single accumulated edge image to accentuate defect-induced thermal disturbances in the form of edge features. Finally, the accumulated edge image is automatically processed using a binary imaging algorithm to visualize the micro defect in the target device. The performance of the proposed reference-free micro defect visualization algorithm is examined using two types of specimens, semiconductor chips and ceramic-epoxy composites. The proposed algorithm successfully diagnoses micro defects ranging from 4 mu m to 40 mu m in width