The transparent area of indium tin oxide conductive film possibly has two kinds of defects, processingtype and handlingtype. According to the characteristics of different defects and requirements from practicable application, an automatic method for defect inspection based on highresolution vision system was proposed. In the system, a coaxial lighting module using Koler configuration was designed, which was suitable for optical property of indium tin oxide film and meets space requirement of working distance 30mm. In addition, the highresolution imaging module applied to practicable inspection was also designed. It shared a doublet lens with lighting module. After obtaining image of indium tin oxide film, two image preprocessing methods were exploited to inspect each kind of defect respectively. If subtracted image processed by median filter in neighboring radius r=7 from original image, scratch defect could be kept clearly. If processed original image by morphological and threshold method successively, transparent circuit pattern could be observed remarkably. The contrast of the pattern image was increased to 48%. The images after being preprocessed provide reliability for automatic defect recognition, and guarantee sensitivity and accuracy to location of indium tin oxide film.