The invention relates to an image processing method for apparent defect detection of a thin-film capacitor. In the process of manufacturing the thin-film capacitor by a manufacturer, defects such as irregularity, burrs, margins, and protruding edges exist on the surface of the capacitor due to process reasons. According to conventional methods, the surface detects of the capacitor are manually detected by sight, so that in the detection process, the labor intensity is great and the efficiency is low. The method provided by the invention comprises the following steps: firstly eliminating unnecessary background by virtue of target prospect extraction; horizontally and vertically projecting to obtain a target capacitor region; and then detecting edge and gradient. The method provided by the invention is low in complexity and fast in operating speed, can be used for satisfying a real-time detection request of a production line and has the advantages of being strong in adaptability and high in detection precision.