18 October 2022 in IEEE Transactions on Image Processing
Recently, graph-based methods have been widely applied to model fitting. However, in these methods, association information is invariably lost when data points and model hypotheses are mapped to the graph domain. In this paper, we propose a novel model fitting method based on co-clustering on bipartite graphs (CBG) to estimate multiple model instances in data contaminated with outliers and noise. Model fitting is reformulated as a bipartite graph partition behavior. Specifically, we use a bipartite graph reduction technique to eliminate some insignificant vertices (outliers and invalid model hypotheses), thereby improving the reliability of the constructed bipartite graph and reducing the computational complexity. We then use a co-clustering algorithm to learn a structured optimal bipartite graph with exact connected components for partitioning that can directly estimate the model instances (i.e., post-processing steps are not required). The proposed method fully utilizes the duality of data points and model hypotheses on bipartite graphs, leading to superior fitting performance. Exhaustive experiments show that the proposed CBG method performs favorably when compared with several state-of-the-art fitting methods.
5 May 2021 in IEEE Transactions on Industrial Electronics
Geometric model fitting has been widely applied in the electronic industry. However, it remains as a challenging task when handling the data corrupted by a large number of false matches (i.e., severe outliers) between two-view images. In this article, we propose a novel motion consistency guided fitting method (MCF) to robustly and efficiently estimate the parameters of model instances in data involving severe outliers. Specifically, from input data, we first generate a series of neighborhood sets, in each of which gross outliers that are inconsistent in motions can be effectively filtered, according to motion consistency among true matches (i.e., inliers). Then, we propose an effective sampling algorithm to sample minimal subsets from the generated neighborhood sets. In this way, the model hypotheses computed from the sampled minimal subsets can cover all model instances with a high probability. Furthermore, by taking advantages of the generated hypotheses and neighborhood sets, we propose a novel model selection algorithm to estimate the number and the parameters of model instances. For fitting evaluation, we also build a new dataset, in which the images are collected from a fundus camera. Experiments on a variety of electronic industrial applications show that the proposed MCF achieves higher fitting accuracy at a much lower computational cost than several state-of-the-art fitting methods.