Freehand sketch based 3D reconstruction using constraints

3D reconstruction from uncalibrated images has many and important applications in computer vision. Different methods exist, but they present various problems. First of all, methods which are full automatic present the problem that they require a difficult and challenging self calibration or calibration step. They need complex calculation operators. And they need more than image to achieve 3D reconstruction. Under all these complicated constraints we don´t have good and reliable results as well as needed. In addition to this, we have the problem of complex used material and wasted memory and time resources. In the other hand, there are interactive methods. They are less heavy than the first ones but, they suffer from the interactivity of user during the whole modelling process. In this paper, we propose a semi-automatic 3D reconstruction approach from a single or more images with less user interaction based on freehand sketches. Basically speaking, we used bundle adjustment based on Levenberg-Marquardt optimization with the aim to estimate our 3D scene. Using extracted constraints from sketches brings us many advantages; it helps us to reduce estimated parameters number and stabilizes good quality for 3D results. We try to proof the importance of freehand sketches use in 3D reconstruction in terms of results reliability and process simplicity.