We developed a method to obtain corner points for healthier point matching using corner properties such as corner angle, corner orientation and contrast. A large corner point set, obtained by a common corner detector (Harris, Tomasi-Kanade etc.) is given to our algorithm as input. Then, the corner properties are extracted for this point set in terns of image derivatives. Cornerness measure calculated from the image is compared with the one calculated using an ideal corner with the extracted properties. If they are close enough, which shows that the neighborhood of the point possess corner properties and estimations are successful, the corner is selected. It is presumed that corners selected by this method are more suitable for point matching. Moreover, extracted corner properties can be used as a priori information for matching.