An Automated System for Image Reconstruction from Distorted Image Fragments

Abstract

Image reconstruction from image fragments has lots of application areas like forensics, art restoration, reconstructive surgery, archeology, puzzle gaming, and civil construction. Reconstruction becomes more challenging when number of fragments increases, image size increases, fragments are similar in shape and color, or fragments are transformed and shuffled. To execute image reconstruction in an effective and efficient manner, an automated computer-based system has been proposed in this paper using image processing techniques. A median filter is first applied to remove noise, after which Speeded Up Robust Features (SURF) are used to orient the fragments properly by reversing any arbitrary transformation applied. Finally, the candidate fragment for each position is chosen using an intelligent fragment selection technique. The proposed method achieves a very good accuracy of 97.78% to 100% and takes minimal time for reconstruction as compared to manual reconstruction.