When using laser triangulation for image acquisition, the surface structure might cause shadows in the laser line due to occlusions caused by the object and triangulation angle. For areas with a shadowed laser line it is not possible to extract the local height information which in turn can be a source of error, as the height image will have holes with no information at those positions and flaws therefore cannot be detected.
To compensate for this systematic deficiency, it is helpful to analyse the same laser line with several cameras at different angles. Any shadows that one camera suffers from, will appear in a different camera's view, so the sum of the combined views will remove any potential for hidden features.
The CVB Merge 3D algorithm consolidates these different images to one height image with dramatically reduced shadow effects. The output image is an intersection of all individual images and finally contains a complete surface with the only remaining holes at positions where all input images were incomplete.
Using CVB Merge 3D in combination with multiple cameras can overcome the systematic deficiency of laser line triangulation and make sure all flaws and defects will be detected in the following inspection.
For customer-specific developments this algorithm has already been used to acquire a full, closed 360 degrees point cloud using multiple cameras and lasers.
The areas that are invisible to either camera are covered by combining the range maps from several cameras to form a merged image. The example below shows the merged range map, and reveals the shadows that are only present on both images.