Image Mosaic Methods Of Brazing Rings Auto-inspection System

Image mosaic is a technology that carries on the image registrationand fusion between two or more images which are acquired in the differentconditions and overlapped with each other, eliminates redundancy ofinformation and generates a high resolution and precision image. Imagemosaic has widely application in the fields of virtual reality, biomedical imageprocessing, video processing and industrial production.In tranditional brazing rings defects detection system, each ring isobtained fully in one image by block image acquisition method. Experimentsshow it can cut off the system prematurely and interfere with the operation ofeliminating the unqualified product. This paper spread the brazing rings out onthe belt which is not need slotted. There may be the same rings or areas,incomplete rings in the overlapping area between the acquired images. Thispaper can get the seamless, high resolution image by digital image mosaictechnology and prepare for brazing rings recognition and detection. Currently,the key algorithms of image mosaic are based on gray information, featuresand transform domain. The algorithm of image mosaic based on the featuresnot only gives attention to mosaic speed and precision of mosaic results, butalso it can overcome the influence of geometry distortion, a small mount ofdistinctive features, noise and fuzzy between two images.In view of the above a few problems, the main contributions and featuresare listed as follows:Firstly, the status of brazing rings inspection, application anddevelopment trend of image mosaic technology are briefly introduced,especially the fundamentals, construction and key algorithm of image mosaictechnology are introduced in details.Secondly, on the particularity of less information, single feature andcomplex type of transformation between images, feature detection anddescription algorithm are analyzed and compared by the flow of imageregistration which is based on local invariant feature.Thirdly, according to the study and research on very famous SIFT andSURF algorithm in recent years, an optimized SURF algorithm is proposed. Experiments show that the new algorithm is much better than SIFT incomprehensive properties, especially SURF speed3~5times faster than SIFT.Finally, aimed to the difference of gray value, image seam, imageblurring after image interpolating and re-sampling, an automatic weightedaverage image fusion optimization algorithm is proposed. When to analysisand evaluate the fused image, the method must be comprehensive, combiningobjective with subjective evaluation standards. Experiments show that thismethod can achieve a better results and obtain a higher quality image withhigh resolution and definition.This paper complete the image mosaic in brazing rings detection system,get the seamless, high resolution images and improve the adaptation of thisauto-inspection system.