An algorithm for automatic recognition of overwater bridge target based on “joint feather and knowledge rule-based” is presented for the problem concerning automatic recognition of overwater bridge target in optical remote sensing images. Firstly, based on knowledge feathers of overwater bridge target, waters in an optical remote sensing image are extracted to narrow down bridge detection range. After preliminary segmentation of waters, mathematical morphological algorithms are used to remove small noise points. Secondly, binary image connected regions are labeled based on pixels to achieve regional features of waters effectively. After the determination of river areas, the area tracing algorithm is adopted to extract outer boundaries of river areas. Candidate areas with possible existing bridges can be prejudged according to bridge knowledge rule. Straight line fitting is applied on points of two boundaries of possible bridge within candidate areas to achieve two bridge boundary lines. It is indicated by analysis that the algorithm raised in this paper shows favorable recognition speed and accuracy.
[1]
Konstantinos Kalpakis,et al.
Fast Algorithms to Implement N-FINDR for Hyperspectral Endmember Extraction
,
2010,
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.
[2]
Rafael Grompone von Gioi,et al.
LSD: A Fast Line Segment Detector with a False Detection Control
,
2010,
IEEE Transactions on Pattern Analysis and Machine Intelligence.
[3]
Bo Zhang,et al.
Knowledge-based bridge recognition in high resolution optical imagery
,
2006
.
[4]
Chen Ai-ju.
Algorithm for Recognizing Bridges in Large Area Satellite Remote Sensing Imagery
,
2014
.
[5]
Antonio J. Plaza,et al.
Parallel Hyperspectral Image and Signal Processing [Applications Corner]
,
2011,
IEEE Signal Processing Magazine.
[6]
Xu Peng-fei.
Novel Algorithm for Recognition of Bridge in Remote Sensing Image
,
2011
.