Space-variant Colored Point Cloud Integration with Reduced Blind Spots

This paper aims to generate space-variant colored point clouds that are efficient for data storage and real-time processing, targeting observation subjects ranging from complex-shaped objects prone to blind spots to wide-area spaces. First, the author describes a local space-variant colored point cloud measurement device and proposes a method for integrating multiple local space-variant colored point clouds measured at different locations using this device. The proposed method is characterized by two key features: first, integrating the local colored point clouds while maintaining their space-variant state to accelerate processing; and second, employing an integration method that avoids unnecessarily increasing the point cloud data volume. The resulting global space-variant colored point cloud exhibits non-uniform point density based on importance and avoids excessive data growth. Consequently, the data storage burden decreases as the observation area expands or the shape becomes more complex. Experimental results verified that the proposed method achieved sufficient accuracy in integrating local space-variant colored point clouds. They also demonstrated the data reduction rate during integration corresponding to the thinning rate of point clouds outside Region of Interest (ROI) determined by the saliency map.