Improved mean filtering algorithm in THz imaging

The de-noising algorithms applied in passive THz imaging was studied in this paper. An improved mean filtering algorithm was proposed, in which the threshold was introduced based on the traditional mean filtering algorithm. The shape of filter mask was decided by the characteristics of scan images resulting from passive THz imaging, and furthermore the best size of the kernel was determined according to one of the evaluation parameters of the digital images-entropy. The experimental result shows that the novel de-noising algorithm has good effects on passive THz images. It not only preserves the information of imaging targets, but also removes out the background noises. The algorithm enhances the quality of the images and increases the ability to detect the concealed metal objects underneath people's clothing. Therefore, it provides necessary guarantees for the application of passive THz imaging technology in the future security inspection.