Case-based object recognition for airborne fungi recognition

Model-based object recognition is a well-known task in Computer Vision. Usually, one object that can be generalized by a model should be detected in an image based on this model. Biomedical applications have the special quality that one object can have a great variation in appearance. Therefore the appearance of this object cannot be generalized by one model. A set of cases of the appearance of this object (sometimes 50 cases or more) is necessary to detect this object in an image. The recognition method is rather case-based object recognition than model-based object recognition. Case-based object recognition is a challenging task.

It puts special requirements to the similarity measure and needs a matching algorithm that can work fast in a large number of cases. It also needs a case acquisition procedure that can capture the great variation in appearance of an object and generalize these data into a case description. In this paper we describe the chosen case representation, the similarity measure and the matching as well as the case acquisition procedure. We evaluate our method based on a large enough set of digital images containing biological objects such as fungi spores.

We can show that the similarity measure is superior to detect the objects in the images. The developed method for case acquisition and learning of generalized cases allows us to learn interactively a sufficient number of cases that are further stored into our case base. Finally, we give results on the performance of the system by calculating the recognition rate.

These result show that we have developed a novel similarity measure for object detection in digital grey-level images and a novel procedure for case acquisition and learning that allows us to learn a sufficiently large enough case base and to generalize over a group of cases.