Multi-attribute smooth graph convolutional network for multispectral points classification

Multispectral points, as a new data source containing both spectrum and spatial geometry, opens the door to three-dimensional(3D) land cover classification at a finer scale. In this paper, we model the multispectral points as a graph and propose a multiattribute smooth graph convolutional network(Ma SGCN) for multispectral points classification. We construct the spatial graph,spectral graph, and geometric-spectral graph respectively to mine patterns in spectral, spatial, and geometric-spectral domains.Then, the multispectral points graph is generated by combining the spatial, spectral, and geometric-spectral graphs. Moreover,dimensionality features and spectrums are introduced to screen the appropriate connection points for constructing the spatial graph. For remote sensing scene classification tasks, it is usually desirable to make the classification map relatively smooth and avoid salt and pepper noise. A heat operator is then introduced to enhance the low-frequency filters and enforce the smoothness in the graph signal. Considering that different land covers have different scale characteristics, we use multiple scales instead of the single scale when leveraging heat operator on graph convolution. The experimental results on two real multispectral points data sets demonstrate the superiority of the proposed Ma SGCN to several state-of-the-art methods.

Weight-Edge Convolution Neural Network for Point Clouds Learning

As a kind of flexible three-dimensional geometric data, point clouds can accomplish many challenging tasks so long as the rich information in the geometric topology architecture can be deeply analyzed. On account of that point cloud data is sparse, disordered and rotation-invariant, the success of convolutional neural network in 2 D image cannot be directly reproduced on point cloud. In this paper, we propose WECNN, namely, Weight-Edge Convolution Neural Network, which has an excellent ability to utilize local structural features. As the core of WECNN, a novel convolution operator called WEConv tries to capture structural features by constructing a fixed number of directed graphs and extracting the edge information of the graph to further analyze the local regions of point cloud. Moreover, a weight function is designed for different tasks to assign weights to the edges, so that feature extractions on the edges can be more fine-grained and robust. WECNN gets overall accuracy of 93.8% and mean class accuracy of 91.6% on Model Net40 dataset. At the same time, it gets a mean Io U of 85.5% on Shape Net Part dataset. Results of extensive experiments show that our WECNN outperforms other classification and segmentation approaches on challenging benchmarks.