cwi-dis/pointpca

PointPCA is a full-reference point cloud objective quality metric that relies on statistical features computed from geometric and textural descriptors. The point clouds first pass through a duplicate point merging step, before computing a correspondence function that identifies matching points between the reference and the point cloud under evalution. Geometric and textural descriptors are then computed, and corresponding statistical features are estimated. The latter are compared to produce predictors of visual quality. A final quality score is obtained by fusing individual quality predictors using a regression method.