Developed Technique for Automatic Point Cloud Texturing Using Multi Images Applied to a Complex Site

The reconstruction of the textured 3D models seems to be very important in almost all the industrial and the engineering fields. These models will aid in defining the problems and consequently finding suitable solutions. In some cases, the final solution could be also simulated on the computer before the applying on the real site. Recently, the laser scanning technology is widely used in capturing the object geometry, whereas the texture is extracted from digital images which are parallel taken to the site. The images are captured either by a camera attached to the laser scanner itself or by a free hand camera. The camera attached to the laser scanner will limit the wide vertical field of view of the laser scanner which exceed the 300 degree, further more the best color quality will be obtained at the best lighting conditions of the image which may not be at the same position of the laser scanner. This is why the free camera positions registered to the laser point cloud approach is still having the potential. The challenge in this approach is how to deal with the occlusions, which normally causes wrong color on geometric details in case of blind texturing. In this paper, a new technique is developed for the texture extraction from multi images applied to the free camera positions registered to the point cloud approach. This technique uses the computer intelligence to automatically detect the occlusions and differentiates between the various planes in the 3D model, which enabling correct texture extraction process. Details on the developed technique together with the approach execution on a real engineering application will be presented. As a part of this work, a special software for the color extraction considering the technique of the occlusions detection is also developed. The final result is a textured 3D point cloud represents the real site. * Corresponding author.