A character defect inspection method based on skeleton template matching is presented to deal with the detection problems of character breakage or stroke break. The dominant color is extracted by coarse segmentation and fine segmentation of the typical pixels from standard image samples. Regional geometric location of connected component is analyzed and text area is located on color planes of each dominant color. The text skeleton template is obtained by using the progressive refinement of color planes in text area. The sample images for inspection to skeleton template are matched in order to gain defect information and the defect information on sample images is labeled. Experiment results show that this method is able to eliminate the effects which are caused by twitter from interferences in the printing process and registration accuracy of text defect detection. It achieves an accurate defect inspection result.