Directional Image Filter Using Rose Curve In Frequency Domain

Feature extraction process for an image is core stage of any application that deals with images. Thus getting different features of a given image as many as possible is an important procedure. One of the most used methods for feature extraction is filter bank. Given image is initially masked with filters that exist in the filter bank and then fusion algorithms are employed. Filters in filter bank are generally pre-prepared and constant filters. Therefore, the constant filters may not capture all features of the given image. In this work, a novel filter that uses rose curves in frequency domain was proposed. Suggested filter has three totally adjustable parameters. When the parameters of the developed filter is changed, different features of the given image are obtained. Consequently, the image filter devised can be used as filter bank for feature extraction process.