Markerless registration of image and laser scan data

A computer-implemented method performed by one or more processors to automatically register one or more first 2D images of a surrounding environment, wherein the computer-implemented method comprises:

• accessing a 3D scan that records at least a portion of the surrounding environment, the portion is also captured by a first 2D image from the one or more first 2D images, and the 3D scan comprises a point cloud and a second 2D image;
• detecting 2D features in the second 2D image from the 3D scan, and associating 3D coordinates from the point cloud to the 2D features in the second 2D image;
• detecting 2D features in the first 2D image from the first type of measurement device;
• identifying matching 2D features from the first 2D image and the second 2D image from the 3D scan;
• based on determining at least a predetermined number of matching 2D features from the first 2D image and the second 2D image, calculating a position and orientation of the first 2D image in a coordinate system of the 3D scan using the matching 2D features.

A system comprising:

• a first type of measurement device that captures first 2D images of a surrounding environment;
• a second type of measurement device that captures at least a first 3D scan and a second 3D scan of the surrounding environment, the first 3D scan captured from a first position and the second 3D scan captured from a second position;
• and
• one or more processors configured to perform a computer-implemented method to register the first 3D scan and the second 3D scan, each 3D scan comprises a point cloud and a second 2D image, wherein the computer-implemented method comprises:
• accessing one or more first 2D images from the first type of measurement device, the one or more first 2D images record portions of the surrounding environment overlapping with the first 3D scan and the second 3D scan;
• generating one or more first localized images by calculating a first pose of the one or more first 2D images with respect to the first 3D scan;
• generating one or more second localized images by calculating a second pose of the one or more first 2D images with respect to the second 3D scan;
• and
• computing a transformation between the first 3D scan and the second 3D scan based on the first pose and the second pose.