Aiming at the problems of difficult, low efficiency, and limited scope of manual bridge disease inspection operations, an image stitching algorithm based on grid deformation alignment and iterative optimization of suture lines is proposed by combining the structural features of bridges and the field of view constraints of inspection robots. In the image registration step, mesh deformation is used instead of the homography transform to improve the local alignment capability. Furthermore, by introducing line segment features based on point features, it mitigates the distortion of the image caused by grid deformation. In the image fusion step, given how much texture intensity is perceived by the human eye, we introduce structural similarity to quantitatively assess the merit of suture lines, and the adaptive saliency is used as a weight for the smoothness term to solve for the optimal suture line via iterative optimization. The experimental results show that the algorithm proposed in this paper can stitch concrete bridge crack images quickly and accurately, and solve the problems of ghosting, position shift, and color inconsistency in image stitching with high accuracy and stability.