Weight-Adaptive Interval Neighborhood Graph for Characterizing the Difference in Spatial Relationships Between Objects

The spatial relationships between two objects can be characterized by their interval relationship. The difference in spatial relationships can be determined by using an interval neighborhood graph. In this paper, we propose to adapt the weights in an interval neighborhood graph according to the significance of change in spatial relationships. We apply our proposed approach in the application of Chinese handwriting error detection. A Chinese character consists of many strokes that contain certain spatial relationships between them. By using our proposed weight-adaptive interval neighborhood graph, we can detect more handwriting errors with a higher accuracy.