A Linear Scoring Algorithm for Shredded Paper Reconstruction

This paper proposes a method of shredded paper reconstruction using a scoring algorithm. This method consists of four modules, i.e., preprocessing, scoring, searching, and reconstructing. The scoring algorithm is created to predict the location of the next black pixel using the concept of linear regression and linear combination. Three black pixels from three consecutive columns are considered to predict the position of the next black pixel. The experimental results show that the proposed method yields the better performance over directly comparing the edges of two shreds when the suitable threshold is used in the preprocessing step.