Accurate data registration and alignment can be tricky, especially in large or featureless areas, but a few best practices can help. First, strategic target placement is key using a well-distributed spheres or checkerboard targets throughout the scan area can greatly improve alignment. If targets aren’t an option, ensuring at least 30-50% overlap between scans helps the software find common reference points.
For featureless spaces, like long hallways or open rooms, consider adding temporary objects (boxes, tripods, or markers) to create reference points. Also, using cloud-to-cloud registration with a high-precision setting in software like Recap, Cyclone, or Faro Scene can refine alignment. If you notice drift, breaking the project into smaller clusters and registering in stages can improve accuracy.
Post-processing is just as important—always check alignment visually and use software tools to fine-tune errors. Have you tried any of these methods, or are there specific challenges you’re facing?
Subhabrata Roy
Construction Consultant
Tejjy Inc
In this situation, we use two Topcon GLS-2000 scanners, with 1.5" SMR monitoring prisms (0.1mm centring error) and a Trimble SX12, we export the observations to least squares analysis using SALSA, with level and GPS vectors processed through AUSPOS (or OPUS) and their covariance matrices from SINEX files. Instead of measuring occupation height, we use resection and if high vertical accuracy is required, use a differential level to measure occupation relative to a local height datum.
We've found the Topcon GLS requires mini prisms at distances less than 150m, otherwise we get vertical angle errors, the longest prism shot with the GLS we've done was 420mm on a 62mm diameter prism.
This will produce much higher accuracy than cloud to cloud algorithms, for example corrugated wall sheeting scanned on both sides, from positions that have not line of sight and no overlap are perfectly aligned, it's also possible to see where the wall sheeting overlaps.
