An SPCP-based thermographic data analysis method is proposed for defect detection. The proposed method is integrated with a moving-window strategy. Noise and non-uniform backgrounds contained in thermal images are largely removed. Better defect detection results are achieved using the proposed method.