Imajing 3D pro
IMAJING 3D PRO is a set of tools to compute depth maps behind 2D imajbox® images and generate 3D point clouds from a single pass.
It brings automatically a 3rd dimension to single imajbox® surveys
WHAT FOR ?
Imajing software and web tools are dealing with 2D images from imajbox® surveys .
These 2D images are basically thousands of photography taken every 2 to 3 meters of the zone that has been surveyed. Each image is precisely geo-referenced and oriented in the space.
To enable 3D positioning and measurements in images, imajing has first introduced photogrammetric tools.
As the raw data material of the survey is only made of a single view 2D images, the photogrammetric 3D positioning requires to pick the exact same point in two consecutive images.
This tools is highly accurate, but requires a precise operation and two click for a single point.
Without having a LIDAR on board, no depth information where existing.
So imajing decided to combine all its know-how on positioning, image processing and code optimization on graphic cards to create imajing 3D pro.
Imajing 3D Pro includes several algorithm that does bundle adjustment, visual odometry, feature tracking, rectification, correlation, filtering and fusion of depth maps to provide 3 outputs :
Imajing 3D depth maps :
Generates for each image of a survey a depth map. This depth map enable to compute the distance between the camera and the object, feature, element pointed on the image, and therefore to compute instantly a 3D position of the element pointed by the mouse on the image.
These depth maps are used by imajview 4 to enable a 3rd positioning tool (photogrammetry, ground plane, depth maps)
The result is the same comfort as mixed LIDAR+IMAGE surveys, but made from a single image source,
The solution of single image source reduces drastically the volume of data collected, and simplifies at most the collection process. (eg: enables surveys behind a windshield, up to 100km/h with imajbox2).
These depth maps will be supported in all the imajnet® tool chain for web diffusion in Q4 2017
PCD pointcloud :
By fusion of depth maps, imajing 3D pro can generate point clouds in .pcd format.
.pcd point clouds are NOT geo-referenced, but are calibrated and can be used for measurements, CAD or 3D modeling.
This module will be available in September 2017 in standard imajing post-process.
.LAS pointcloud :
.LAS are geo-referenced point clouds that can be used in any 3D point cloud tools for modeling, measurement, geo-referencing points, integrating in a DTM, etc..
This module will be for sale in Q4 2017
Questions and answers about imajing 3D Pro :
o Depends extremely on the hardware configuration you use. Imajing 3D pro is optimized to exploit GPU on NVidia graphic cards. With an average NVidia graphic card count on 7 x the survey time as an estimation.
o However the average computation time can be highly reduced when parallelizing multiple surveys processing. Then the average time of processing per km of survey can be reduced by 4 to 8 or more according to your CPU (core i7 or i9, AMD Rizen) and the RAM available on your graphic card.
o The relative accuracy depends mostly on the distance to the camera. Up to 20 m from the camera, relative accuracy is better than 10 cm DRMS. Accuracy decreases with the distance.
o .LAS absolute accuracy will depend on the positioning solution used.
With a point cloud rebuilt up to 20 m from the camera, the absolute accuracy will be in the 60 cm DRMS with dGNSS positioning solution or 20 cm DRMS with RTK (RTK fixed)
o Imajing 3D Pro can rebuild up to 100m. However, as the error grows with the distance, this is not recommended (except for inaccurate needs such as detecting walls or obstacle).
The typical corridor shall be rebuilt within 20 to 30 m.
20 m is the default reconstruction distance.
o Yes, the main point is image quality (clear, sharp, well contrasted, without dirt or masks) and an inter-distance between images configured to be within 1 to 2,5 m for a good coverage.
o For more details, ask our sales team for « Best practice to obtain good 3D with imajing 3D pro » : (firstname.lastname@example.org)
o No, certainly not. And even if the output looks very similar, LIDAR point cloud is not colorized (or done in post processing with an additional camera) where imajing point clouds are colorized with actual colors directly.
o This image based approach very much depends on images themselves.
If your goal is to map all in 3D with a few centimeter accuracy, you shall go for LIDAR; If your need is simplified use, if you have cost constraints, if your need is availability of 3D on some part of the surveys only in addition to photogrammetry, then imajing 3D pro with imajbox® is the most effective solution.