Abstract: In infrared thermographic non-destructive testing, to reduce the effects of non-uniform heating and increase defect contrast and inspection depths, the thermal image sequence processing algorithms were studied based on the pulsed thermography of carbon fibre reinforced plastic samples. The factors that produce uneven heating were analyzed and summarized. Three types of data processing algorithms, including deleting fitted background, polynomial fitting and integrated technique, were proposed based on the least square method. The algorithms were verified and compared by experiments and their processing effects were quantitatively evaluated by three statistical standards of image evaluation. The result shows that the deleting fitted background technique can effectively remove the bad influence of uneven heating on the defect inspection, the polynomial fitting method can decrease uneven heating effect and increase defect contrast at the same time, and the integrated approach has the advantages of the two forenamed methods and can more greatly enhance defect contrast.