An Image Matching Using Critical-Point Filters and Level Set Analysis

Abstract

Critical-Point Filters (CPFs) are nonlinear filters which preserve intensity and location of each critical point in the image and reduce the resolution without any prior knowledge. Although CPFs can avoid blurred intensity and ambiguous location problem of previous linear filters, its computational cost is still expensive due to its complexity. We propose an enhancement of the CPFs algorithm for image matching using level set analysis. An image is analyzed and transformed to hierarchical level sets of pixel having same intensity. Connectivity of the level sets represents the image contrast invariant features. Between the corresponding level sets of the input images, two pixels are mapped based on their energy and bijectivity conditions. Finally, less computational time with precise mapping is shown in the experimental result.