In the process of equipment production in large manufacturing, the continuity of production is becoming more significant. Timely detection of equipment operation faults can ensure production continuity and greatly reduce loss. In this study, the purpose is to use multi-equipment and image stitching algorithm to obtain the complete image of a large-scale production line. An improved image stitching method based on image fusion is proposed in this paper, which mainly solves the technical problems of stitching seams, unnatural effects, and distortion after image transformation in the existing stitching technique. In the image stitching algorithm, the improved fusion algorithm based on the optimal seam and gradated in and out fusion algorithm is used to realize image fusion, including the use of dynamic programming to find the optimal seam and limit the range of fusion based on the optimal seam found. Finally, the gradated in and out fusion algorithm is used to perform fusion calculation within the limited fusion range to complete image stitching. In the end, through the comparison of different dimensional image fusion indicators with the effect of the existing fusion algorithm, the experimental results show that the method in this paper solves the problem of unnatural image stitching effect, enhances the image stitching result, and has the great fusion effect. Therefore, the panorama processed by the image stitching algorithm proposed in this paper can be efficiently processed through the industrial detection module.