Linear feature such as lines and curves are one of important features in image processing. In this thesis, new method of significant linear feature detection is suggested. Also, we studied approximation technique which transforms detected linear feature into data structure suitable to real application. This method is based on graph theory. The basics of this method is minimal spanning tree, which is used in edge linking process. After that processing, hairs and inconsistent line segments are removed. To approximate and describe traced linear feature, piecewise linear approximation is performed. The algorithm is demonstrated through examples.