Luggage Image Mosaic Of X-RAY Security Device

With the development of society, people pay more attention to the security of public places. X-ray security check device can check out the items in luggage without opening it. To ensure public security, x-ray security check devices are now widely used in public places such as railway station, bus station and airports. It could happen that luggage stops in the channel due to some reasons while scanning. This is so-called scan-interruption. When scan-interruption happens, the image of the luggage is not complete. Image mosaic technology can combine the images into a complete one, which are taken before and after the interruption.The thesis analyzes the construction of the pseudo-dual energy x-ray security check devices and the imaging principle of dual-energy x-ray transmission image. The features of the image are investigated according to the system structure and imaging principle. Normalization process is chosen as the image pre-processing method according to the features and trailing data and redundant data are removed after pre-process. NCC (Normalized Cross-Correlation) based on edge intensity is proposed as the similarity measure of dual-energy x-ray transmission image. This method uses local standard deviation to calculate the edge intensity of each pixel and converts edge intensity to contribution factor, then uses the pixel value and contribution factor to calculate the NCC of low energy image and high energy image respectively, and last calculates the integrated cross correlation from the NCC of low energy image and high energy image. Experimental results show that NCC based on edge intensity improves the accuracy of similarity measure between two dual-energy x-ray transmission images. NCC based on edge intensity also improves the accuracy of image registration.After getting the matched position by image registration, image fusion is needed to fuse the overlapped area. The thesis proposes an image fusion method based on membership. This method first calculates the membership to the image after scan-interrtuption of each pixel, and then converts the membership of each pixel to the weight. This method uses the weight to fuse high energy image and low energy image respectively. Experimental results show that the fused luggage image is complete and has no visible seams on the image.