Research On Feature Extraction And Image Stitching Of Weak Texture Surface

The feature extraction algorithm is not easy to identify because of the weak texture features that have a very small difference in the pixel value.So weak texture surface feature extraction is one of the most important problems in the field of digital close-range measurement and image processing.At present,the method of pasting artificial mark points is often adopted for the feature extraction of the weak texture surface,that the feature points are artificially created on the surface of the weak texture target so as to perform subsequent measurement and the like.However,for the visual measurement of soft texture and high-precision equipment on weak texture surface,it is easy to cause irreversible damage to the target surface.Therefore,a method to solve the problem of surface damage caused by sticker artificial signs is urgently needed..Aiming at this problem,this paper proposes a method of using laser graphics as a pseudo feature point of the texture to be used to replace the sticker artificial mark.It makes full use of the characteristics of laser monochromatic,high brightness and good direction.We transform the output lens of laser to make enough feature points on the surface of the object to be tested.Then the target to pick up the characteristics of the weak texture surface and the image splicing are realized.Finally,the work of image detection of the object to be tested with weak texture is completed.The specific research scheme is as follows:Several different graphics are designed as alternative laser marking points,and projected onto weak texture surfaces.Combined with feature point pickup algorithm,we verify the shape of laser markers and extract algorithm of image feature points.In the two-step registration of the image registration,the NCC algorithm is used for coarse registration,then the RANSAC algorithm is used to accurately register,so as to improve the registration precision,and finally,the image fusion is carried out by using the weighted average method,so that the feature extraction and image splicing work of the texture surface are completed.The method proposed in this paper is compared with the existing method to paste artificial auxiliary sign points,so as to save the stickup work of the manual mark point.And because the laser auxiliary sign point is a non-contact sign point,it will not cause damage to the surface of the object.It can improve the extraction efficiency of the features while ensuring the surface quality of the weak texture.In addition,this paper studies the image stitching method of large and weak texture surface features to be measured.By comparing the effect of image stitching based on two methods of laser aided logo and sticky marker,the effectiveness of image mosaic technology based on laser feature points is verified,and achieved satisfactory results.