Latex folder contains all LaTex source documents of the thesis and PDF folder contains the compiled version of it.
The demand for three-dimensional (3D) virtual models of complex objects has been increasing in many fields. Highly realistic 3D models are daily used to visualize and simulate events in the medical field, games, architecture, film industry, 3D printing, among others. One of the major limitations to the widespread use of these techniques is the high cost of these 3D models as they are elaborated manually. If the object to reconstruct comes from the real world, the process involves planning a set of views, physically altering and manipulating the position of the object and the sensor, capturing the object, recording the acquired geometric data into a common framework, and finally, integrating the acquired information into a virtual, consistent, and non-redundant model. Given the nature of this process, validating the correctness of the generated model involves much time, effort, and money.
This paper presents a fast three-dimensional reconstruction technique using only pairs of stereoscopic views of the object to reconstruct taken with a low-cost camera (webcam). The purpose of this technique is to reduce the cost of the reconstruction process by not requiring specialized equipment or personnel, as well as reducing its duration. The novelty of this technique is that it uses dense pixel tracking algorithms along with a pyramid technique for fast image processing and a rapid triangulation process to generate three-dimensional information. Generally, dense pixel tracking is not suitable for use in rapid reconstruction due to the considerable amount of data to be processed, however the proposed technique allows to obtain high quality results within a few seconds without using any specialized equipment.
computer vision, fast 3D dense reconstruction
https://www.youtube.com/watch?v=KWHvYrKSWlY
https://www.youtube.com/watch?v=QNWGJnPHDVQ
https://www.youtube.com/watch?v=Fswoi420_NQ
https://www.youtube.com/watch?v=udrLyqW9mQw
If you find this document useful buy Randy Saborio a beer.