Combining iterative inverse filter with shock filter for baggage inspection image deblurring

Abstract

In this paper, we describe an image deblurring algorithm for images generated by the baggage inspection system. Baggage inspection images have low-extent blurring, large intensity dependent noise and need line by line processing in real time, which makes most of the existing methods unsuitable. With these special characteristics, we propose a new algorithm by combining the iterative inverse filter and the shock filter. At each iteration of the inverse filter, the constraint borrowed from the shock filter is imposed so that the image is deblurred without ringing artifacts. The algorithm is fairly fast and can process the image line by line, which can satisfy the real-time requirement. It is also easy to program and can be implemented in practice. The algorithm is tested on the synthetic data and real data from the airport. The experiments show that our algorithm has a great improvement on human's perception and is better than the original algorithms. © Springer-Verlag Berlin Heidelberg 2006.