Empty Area Matters: Bivariate K-Nearest Neighbor Based Scatterplot Overlap Removal Algorithm

The overdraw problem in scatterplot visualization can distort key distribution features like KNN relationships, density, and shape, disrupting visualization tasks. Existing techniques focus on understanding occupied region features but neglect relationships involving empty areas. Recent methods using dummy nodes to represent empty area features have been inadequate. This paper proposes a novel approach based on bivariate KNN relationships using angle and distance, and employs similar triangles to displace points to address overdraw. A linear mapping assigns displacements to points outward from the center, linking empty and non-empty areas and preserving overall distribution features. Evaluations confirm the effectiveness of our method in preserving distribution features of empty areas and demonstrate its potential in solving overdraw issues in bubble charts and aiding trajectory analysis.