Abstract
Generic object recognition, which is one of the image recognition, is an important research area in machine vision. In the image recognition, image is treated as a set of local features and is represented by capturing of the distribution on feature space. Fisher Vector is known as a powerful technique for feature encoding, and is captured mean and covariance components. This paper presents a parameterization of Fisher Vector by using Generalized Fuzzy c-Means (GFCM), and image was represented in the same way as Fisher Vector. The effectiveness of our approach was verified by comparing with the recent image representation techniques.
