Abstract: Provided is a method of classifying colors of a color based image code. The colors of each cell forming the color based image code are classified by sampling a predetermined number of pixels from each cell and then applying a max channel based clustering method, a white balance based color clustering method, a single-linkage algorithm based color classification method, a K-means algorithm based color classification method, or the like. Accordingly, the colors of an image code can be accurately recognized despite of a color distortion due to characteristics of a camera, a printing medium, etc.

Abstract: Reference data for automatically inspecting shapes of fillets formed on a substrate are set to an inspection device that illuminates the substrate from specified directions to generate an image. For each type of components mounted to the substrate, a database is prepared, registering sets of reference data corresponding to different fillet shapes in correlation with heights of solder for forming fillets having these shapes. After components to be an object of inspection are identified, specified steps are carried out on each of these components, including the step of obtaining data on the height of solder for forming the fillet related to a land for which a target area for inspection has been set and reading out reference data corresponding to the data obtained from the reference data registered in the database.