Font Size: a A A

Development Image Stitching System And Research On Crack Extraction Method

Posted on:2018-09-11Degree:MasterType:Thesis
Country:ChinaCandidate:Y Y WeiFull Text:PDF
GTID:2428330542484220Subject:Engineering
Abstract/Summary:PDF Full Text Request
The wind-power blades will be cracked under the effect of long-term external,so its health monitoring and maintenance have become very important.But manual method has low efficiency and security risk,we monitor it by climbing robot carrying image acquisition device.Due to the limitations of photographic equipment,ordinary camera can only collect small scale.Therefore,the image stitching technology is used to match a set of images that overlap each other,and then become a clear image with a large field of view.In this paper,the basic theory and key technologies of image mosaic are studied systematically,and the crack joint system is designed.First of all,about the crack image stitching.Due to crack detection accuracy and time requirements,the paper puts forward a optimization algorithm to prevent Harris corner leakage and surrounded which improves the Harris algorithm to extract feature initiatly,and combines Scale Invariant Feature Transform algorithm to make it have the scale invariance,and uses the algorithm of Best Bin First and do an exact match by Random Sample Consensus algorithm to estimate the transformation matrix at last.In order to prevent multiple image splicing cumulative error,the paper uses the method of bundle adjustment optimization;Finally using the weighted average method to fuse image by image fusion technology.The experimental results demonstrated that the crack image stitching method had high matching precision,strong robustness.Secondly,in order to guarantee the precision of crack detection,the adaptive threshold image edge detection based on the improved wavelet transformation and Morphological repair is presented,which can keeps the crack shape and size identically.The experiments shows that the algorithm is stable and good extraction.After repairing the crack,the paper extracts the crack skeleton,uses projection features to judge crack types,and calculates the crack length,width and area finally.Thirdly,the crack stitching system of the robot.The system is programmed based on VC++ platform,which includes the video decoding,features extracting,features matching,image fusion and crack length calculating and the final image output.The software can play video automatically and complete image mosaic which has the advantages of reliable and a certain practical value.
Keywords/Search Tags:Feature Extraction, Image Mosaic, Harris Algorithm, The Wavelet Transform
PDF Full Text Request
Related items