Specular reflection appears with metal, plastic, glass and many other solid objects. Therefore it is important to establish a technique to recognize specular objects. But Recognizing shiny objects with specular reflections is a hard problem for computer vision. Specular reflections apperar, disappear, or change their shapes abruptly, due to tiny movements of the view. Traditionally, such specular reflections are discarded as annoying noise for recognition purposes.
To get over such difficurties and to recognize specular objects easily we employed ring illumination system in which a ring source is positioned on the axis of the camera. If specular objects are illuminated by the ring they show their own distinctive specularity features in accordance with their shapes. These specularity features represent a portion of iso-inclination on the object surface from which we can infer the shape of the object.
In this paper we perform two recognition tasks using these specularity features by ring illumination. One is to estimate shape parameters like diameter or length, and positions in 3-D space of cylinder, corn, sphere and plane. We can do this by deriving the relation between the specularity features of iso-inclination and the shape parameters or positions. The other is to extract the shapes of general objects from a series of specularity features obtained by scanning ring illumination sources. To finding out the relation between the successive specularity features makes this task possible.
At first we will derive the relations above and then perform a series of experiments. The experiment results show the performance for these tasks by using this method.
