....
Here the data have been cleaned to remove all data (trees, vegetation, and a tractor) that are not associated with the feature of interest. This data is part of the Jacobs Rock data set.

Once all of the scans are cleaned, they are aligned in the IMAlign module in Polyworks. Scan alignment is the process of piecing together or merging together scans to form a complete model of the structure. Scan alignment is performed by identifying common reference points in each of the scans and then co-registering the scans to a common coordinate system. Registration targets were not used in this project. Instead, natural features in the rock (i.e. an indentation in the rock, the tip of a petroglyph, etc…) were used for scan alignment.

Once all of the scans for a site are aligned, the data are merged into a polygonal mesh file. Creating the polygonal mesh file consists of converting the data from a mass point cloud to a polygon mesh which contains connectivity information between the points. A polygonal mesh file must be generated in order to transfer the data between software applications .

Before the data can be exported from Polyworks, it must undergo additional editing. The data will often contain “holes” caused by features such as vegetation that obstruct the scanning of the rock face. While the vegetation data is removed during the initial cleaning process, the “hole” in the rock caused by the vegetation must be filled in. Holes in the data are filled in using a combination of automatic and manual hole-filling algorithms available in the IMEdit module of Polyworks.

Once all or most of the holes in the data are filled in, the final model is smoothed and reduced to prepare it for export. Mild smoothing is performed on each rock model to filter out the natural undulations or “noise” in the laser scanner data. The smoothing process improves the overall appearance of the model without greatly reducing the model's accuracy. The data are additionally reduced to make the model easier to handle during the visualization process. The size of the initial data models varies anywhere between 1 – 2.5 million data points. These files are simply too large to import into SoftImage. Therefore all of the data models are reduced to approximately ½ million data points using a surface-based reduction algorithm available in Polyworks. This algorithm filters out data points while preserving the inherent shape of the object. Finally, the reduced model is exported from Polyworks and imported into SoftImage.

.......
Original Narrows data set (left) compared to final Narrows 3D polygonal model (edited, smoothed, and reduced)