Paper Title
Improved Feature Extraction Using Box Filter Combination In Image Data

Abstract
A feature of the image is the point that can represent better the characteristic of the image other than points. Therefore, a feature of the image data must be detected regardless of the change of scale, rotate and viewpoint, etc. For feature extraction of existing research, feature point is extracted compare to normal image and abnormal image, otherwise method has been developed to extract edge points. In particular, SIFT(Scale Invariant Feature Transform) applies the filter and uses DoG(Difference of Gaussian)to extract the edge points. In this case, Gaussian filter is used. The reason is known that the best filter to extract the feature of image is Gaussian filter. However, Gaussian filter blur objects in image. So, using Gaussian filter prevents to extraction exact feature point. Therefore, the purpose of this study is to solve the problem by developing a filter with linear combination of box filters. The proposed filter extracts feature that located the boundary between the object in image more sensitive. In addition, through experiments,the proposed filter is confirmed to be better than Gaussian filter in specific image. Keywords� Gaussian filter,box filter, SIFT,Combination of box filters.