A METHOD AND APPARATUS FOR CONTOUR-BASED OBJECT CATEGORY RECOGNITION ROBUST TO VIEWPOINT CHANGES

摘要

PURPOSE: A method for recognizing a boundary object based on an outline strong against a view point change of an object and a device thereof are provided to strongly operate against a view point change of the object without assuming a view point of the object in advance. CONSTITUTION: A data Input unit(100) receives image data. A category object recognizing unit(200) recognizes a category object among the image data. An output unit(300) outputs a recognition result of the category object recognizing unit. The category object recognizing unit includes an outline detecting unit, a feature extracting unit, a spectrum matching unit, a matching result improving unit, and a matching location predicting unit. The matching location predicting unit predicts a matching location in the image data. The matching location predicting unit recognizes an object in a category.