Four-dimensional augmented reality models for interactive visualization and automated construction progress monitoring

A computer-implemented method comprising:

• storing in a non-transitory computer-readable storage medium multiple, unordered images obtained from photographs taken at a site and instructions executable by at least one processor;
• melding, using at least one processor, the multiple images to reconstruct dense three-dimensional (3D) and four-dimensional (4D) as-built point cloud models (“as built models”) comprising merged pixels from the multiple images in 3D space of the site, where time is a fourth dimension added to the three dimensions of the site captured from the unordered group of images;
• integrating the merged pixels of the as-built models with pixels of the as-planned model to generate an integrated scene;
• transforming the 3D as-built model to a site coordinate system existing within a 3D as-planned building information model (“as-planned model”);
• applying, using at least one processor, a probabilistic model that traverses the integrated scene and labels different areas of the integrated scene according to as-built occupancy and as-planned visibility, comprising:
• assigning each of a plurality of voxels a set of labels including as-built and as-planned labels based on application of the probabilistic model;
• and
• assigning a state to respective voxels with as-built labels based on a consistent visibility from projections of respective voxels onto the unordered group of images, the consistent visibility being between the merged pixels of respective voxels and a plurality of camera locations of cameras that provide the unordered group of images;
• where assigning the set of labels and the state to respective voxels is performed in an order of increasing distance from the plurality of camera locations.

A non-transitory computer-readable storage medium comprising a set of instructions for configuring a construction progress monitoring system executable by a computing device having a processor and memory, the computer-readable medium comprising instructions to direct the processor to:

• store in the memory multiple unordered images obtained from photographs taken at a site in response to user inputs requesting to store the photographs;
• meld the multiple images together to reconstruct dense three-dimensional (3D) and four-dimensional (4D) as-built point cloud models (“as built-models”) comprising merged pixels from the multiple images in 3D space of the site, where time is a fourth dimension added to the three dimensions of the site captured from the unordered group of images;
• transform the 3D as-built model to a site coordinate system existing within a 3D as-planned building information model (“as-planned model”);
• integrate the merged pixels of the as-built models with pixels of the as-planned model to generate an integrated scene;
• and
• apply a probabilistic model that traverses the integrated scene and labels different areas of the integrated scene according to as-built occupancy and as-planned visibility, wherein to cause the at least one processor to apply a probabilistic model comprises to cause the at least one processor to:
• assign each of a plurality of voxels a set of labels including as-built and as-planned labels based on application of the probabilistic model;
• assign a state to respective voxels with as-built labels based on a consistent visibility from projections of respective voxels onto the unordered group if images, the consistent visibility being between the merged pixels of respective voxels and a plurality of camera locations of cameras that provide the unordered group of images;
• where to assign the set of labels and the state to respective voxels is performed in an order of increasing distance from the plurality of camera locations.