The paper presents an efficient spectral method for finding consistent correspondences between two sets of features, which is applicable to various computer vision tasks such as object recognition, shape matching, and registration. The method constructs a graph where nodes represent potential correspondences and edge weights reflect pairwise agreements. Correct correspondences form strongly connected clusters, while incorrect ones do not. The solution is recovered by identifying the main cluster using the principal eigenvector of the adjacency matrix and applying mapping constraints. The approach is robust to outliers and noise, faster than existing methods, and can handle large datasets. Experimental results demonstrate its effectiveness in various scenarios, including 2D point matching, non-rigid deformations, and object recognition from low-resolution images.The paper presents an efficient spectral method for finding consistent correspondences between two sets of features, which is applicable to various computer vision tasks such as object recognition, shape matching, and registration. The method constructs a graph where nodes represent potential correspondences and edge weights reflect pairwise agreements. Correct correspondences form strongly connected clusters, while incorrect ones do not. The solution is recovered by identifying the main cluster using the principal eigenvector of the adjacency matrix and applying mapping constraints. The approach is robust to outliers and noise, faster than existing methods, and can handle large datasets. Experimental results demonstrate its effectiveness in various scenarios, including 2D point matching, non-rigid deformations, and object recognition from low-resolution images.