Straight Line Extraction by Heuristic Search for SAR Image

Straight line feature is the basis of linear target recognition. The existing heuristic search algorithms are suitable for straight line feature extraction in optical image while performing poor in SAR image. For this problem, a straight line extraction method for SAR image based on heuristic search is presented. Firstly, SAR image is filtered by Frost filter. Then, edges are detected by ratio of exponential weighted average (ROEWA) operator and non-maxima suppression location algorithm is adopted to obtain the binary edge map. Finally, considering the local information sufficiently, the strategy of search twice and the principle of “go straight” is utilized and the cost function and rules of heuristic search are designed. Then the straight line extraction is completed. The experiment results show good anti-noise performance and fracture resistance of the proposed method which can effectively describe the straight line feature of SAR images.