Abstract: It is necessary to periodically inspect the blade surface of the wind turbines operation. The widely used manual inspection method brings a heavy workload

high cost and low detection rate

so it is urgent to need anautomatic inspection system.Image mosaic of wind power blades is a key part of the intelligent inspection system

which directly determines the system performance. Current splicing methods cannot extract enough feature points when facing a strip object with a single structure and color such as wind power blade

causing failure in splicing success. A splicing method based on line features is proposed to solve this problem. This method first detects straight lines in the images of the wind power blade

then selects the deduplication as the feature for image registration

and finally splices the blade image according to translation rotation matrix.Experimental results show that the proposed method shows strong robustness and stability in complex situations such as single structure

small coincidence degree and diverse backgroundstability.