This paper proposes a novel and effective model fitting method to estimate multiple geometric structures on a two-layer network, where vertices in the first layer denote model hypotheses and vertices in the second layer denote data points. Instead of only considering the consensus information on model hypotheses or the preference information on data points, we combine these two kinds of information into a two-layer network. Based on this formulation, we first distinguish vertices with the quantities of information they contain in both layers by using an information theoretic algorithm. We then fuse the retained model hypotheses in the first layer together with the generated hypotheses from the retained data points in the second layer. Finally, the proposed method, namely Information Fusion on Two-Layer Network (IFTLN), detects model instances from the fused model hypotheses vertices according to three key elements (i.e., the local maximum value of weighting score, the distance between vertices, and the local density). Overall, IFTLN can not only automatically and simultaneously estimate the number and parameters of model instances with a large number of outliers, but also effectively handle significantly unbalanced distribution of data points among model instances. Comprehensive experiments are performed on both synthetic data and real images, and superior performances are achieved by the proposed method in comparison with some state-of-the-art model fitting methods. (C) 2020 Elsevier Inc. All rights reserved.

In this paper, we present a new image dehazing method via fast multi-scale image fusion. It is designed based on a fusion strategy. Instead of estimating the exact global atmospheric and the transmission separately as most previous methods did, our method directly constructs initial dehazing images with different exposure through the histogram analysis and L0 gradient minimization with adaptive boundary constraint to improve the visual dehazing effect. Experimental results show that this method outperforms state-of-the-art haze removal methods in terms of both efficiency and the dehazing visual effect.

This paper proposes a robust model fitting method, called Outliers Removed via Spectral Clustering (ORSC), to estimate multiple inlier structures in the presence of a large number of outliers. The basic idea is to cast each data point to the conceptual space, where the distance distribution of inliers and outliers from the origin is significantly different. Therefore, all the points can be classified into inliers and outliers according to the distribution of points of each subspace, which is obtained by a spectral clustering algorithm. Furthermore, we can use the clustering result to guide the follow-up sampling to get more clean data points for hypotheses generation when handling the complex multi-structure model with a large proportion of outliers. Experimental results show that our method achieves superior performance compared with some state-of-the-art methods in terms of line fitting, circle fitting, and real images.

It’s still a challenge to recognize object with RGB-D information. HMP is a classical method based on sparse coding, which can adapt to learn feature from RGB-D. HMP method ignore gradient information. And SIFT based sparse coding could capture gradient information well, while cannot adaptively extract other feature from RGB-D. So we propose multi-feature joint sparse representation (MJSR) algorithms, which combine sparse coding based on SIFT and HMP. At first, we extract dense-SIFT from image. Then dictionary is captured with K-SVD algorithm. Sparse coding can be obtained by using Matching Pursuit (MP) on dictionary and sift features. Spatial pyramid pooling is applied on sparse coding based on SIFT and the features consist of patch feature and associated sparse coding as HMP to capture image feature. In the end, we conduct experiment on Washington RGB-D object dataset.

Hierarchical matching pursuit (HMP) is a popular feature learning method for RGB-D object recognition. However, the feature representation with only one dictionary for RGB channels in HMP does not capture sufficient visual information. In this paper, we propose multi-channel feature dictionaries based feature learning method for RGB-D object recognition. The process of feature extraction in the proposed method consists of two layers. The K-SVD algorithm is used to learn dictionaries in sparse coding of these two layers. In the first-layer, we obtain features by performing max pooling on sparse codes of pixels in a cell. And the obtained features of cells in a patch are concatenated to generate patch jointly features. Then, patch jointly features in the first-layer are used to learn the dictionary and sparse codes in the second-layer. Finally, spatial pyramid pooling can be applied to the patch jointly features of any layer to generate the final object features in our method. Experimental results show that our method with first or second-layer features can obtain a comparable or better performance than some published state-of-the-art methods.