Search

Road recognition from SAR image is a very important problem. In this paper, a new method is proposed which can detect multiple roads in an image. Firstly binary edge image is gained by Canny operator and edge refining. Secondly, edge curves are encoded by reversible chained list, and fitness function is constructed based on SAR image road models, and optimal road boundary curves are linked by improved genetic algorithm. Thirdly, complete road boundaries are detected by curve fitting. At last, mated long road edge curves are found out and road surface are stuffed. The experiment shows that this algorithm can realize roads recognition and eliminate influence of other non-road.

Stereo matching with regions is a master method for binocular image matching of wide baseline. It is most difficult that extracting invariant region. Stable extremal region(SER) is proposed in this paper, and its extracting algorithm. To magnify the intensity change and identify its edge easily, the exponent transform has been used. Then, the affine invariant moments of SER and the Euclid-distance between the region and another are computed using in measuring similarity. The experiment result show the extremal points locate accurately, the SER are stable, and correspondence regions are exact, it is improved to 70.9%.

Road recognition from SAR image is a very important problem. In this paper, a new method is proposed which can detect multiple roads in an image. Firstly binary edge image is gained by Canny operator and edge refining. Secondly, edge curves are encoded by reversible chained list, and fitness function is constructed based on SAR image road models, and optimal road boundary curves are linked by improved genetic algorithm. Thirdly, complete road boundaries are detected by curve fitting. At last, mated long road edge curves are found out and road surface are stuffed. The experiment shows that this algorithm can realize roads recognition and eliminate influence of other non-road.