Research And Application Of Multi-focus Core Image Fusion

With the development of computer technology and industrial cameras,core image scanning technology has gradually become an important means of rock and mineral analysis.It realizes the conversion from solid cores to permanent high fidelity digital core images,which is of great significance for the development of petroleum exploration and geological research.Due to the focal length limitation of the optical imaging system during the scanning process of rock mass,the images captured in the defocused area often have varying degrees of blurring,which means that it is impossible to obtain a complete and clear core image.The texture and fractures of rock mass contain important geological information,and the blurring of details has brought great inconvenience to scientific research and observation.Therefore,a multi focus image fusion algorithm（MFIF）is needed to fuse multiple core images with complementary depth of field into a fully clear image^[1],eliminating information redundancy,and improving local clarity and contrast.As an important research direction in the field of image fusion,MFIF algorithm provides an effective solution to the problem of focal length constraints in optical imaging systems.This allows images with different focal lengths in the same scene to be captured in any detail through fusion technology,and improves the resolution and overall information load of the area.At the same time,this technology can also well locate spatial depth,providing good theoretical significance for research in three-dimensional reconstruction,lens calibration,and other directions.Currently,multi focus image fusion algorithms have been widely used in the fields of post photography,medical imaging,remote sensing information,military detection,and so on..MFIF algorithm has been widely studied and applied since its development.Among the relevant algorithms proposed by many scholars,according to their principles,they can be mainly divided into three categories:spatial domain,transformation domain,and deep learning methods.However,existing image fusion algorithms still have shortcomings that are difficult to eliminate.For example,the imaging distortion caused by wide-angle lenses,and the shortcomings of various algorithms in terms of noise resistance,image continuity,and time complexity.Therefore,in order to improve such problems,this paper proposes a general optimization method based on image preprocessing and post processing,as well as a decision graph generation algorithm with higher segmentation accuracy,in order to solve the common phenomena such as block effects and artifact contours in traditional algorithms.The main research content of this article includes the following three points:1.Aiming at the problems of limited object distance difference,complex deformation,and distortion in three-dimensional scanned images,a preprocessing method based on triangulation and affine transformation is proposed to optimize the accuracy of the source image in terms of scale uniformity,ensuring the registration accuracy of the fused image.2.An image post processing method aiming at the portability of the focused segmentation MFIF algorithm is proposed.Using spatial domain operators to extract clear components,spectral consistency is achieved through Poisson fusion during the stitching process;At the same time,pseudo edge suppression is performed at edge fusion points through discriminant functions to improve visual perception and eliminate information redundancy,in order to meet the application requirements of higher fusion accuracy.3.An image fusion method based on NSML focus segmentation and multidimensional adaptive filtering is proposed.Firstly,this paper adjusts the weight of the kernel function of the Laplace operator on the image to better adapt to the detection of multi focus images.At the same time,a Laplace matrix with an optimized kernel function is constructed for the unified scale source image,and a multidimensional adaptive kernel function with segmented scales is designed to purify the matrix;And performing focus judgment on the source image according to the final decision map to obtain a purified full-focus image;Finally,the algorithm and the optimization algorithm are comprehensively compared through two sets of experiments.The results show that this method not only has good operational efficiency,clear visual perception,but also has better data results in terms of evaluation factors.The multi focus fusion algorithm proposed in this paper simplifies the process of multi scale decomposition,and makes reasonable improvements to the construction of the Laplace matrix;In mask purification,a multi-dimensional adaptive filter is proposed to replace the traditional consistency check method,which greatly improves the focusing accuracy and anti noise performance while ensuring operational efficiency.In terms of algorithm innovation,a fusion optimization method of image preprocessing and post processing is introduced to solve the problems of ghost,ghost,and block effects generated by the source image in areas with similar sharpness,and improve image continuity.It provides theoretical significance and reference for the application of image fusion algorithms in other fields.