Shape equalization of binary images using Fourier descriptor interpolation in low-level processing for image understanding

One of the most essential tasks in the image understanding systems is shape analysis of the binary images. The effectiveness of the shape analysis process requires an accurate description of the shape of the object and an appropriate structure of the input data. In this paper, a new method of the shape equalization is presented. The proposed approach increases the accuracy of the shape description without increasing the complexity of the input information. The method is based on the data interpolation using Fourier descriptors and takes advantage of the statistical properties of the objects found in image scenes. A 1-D filter is used for spatial filtering of the contours.