Wave Radiation Image Fusion Recognition Algorithm

Because of good transparency of the clothing at millimeter-wave band, the identification and location of contraband (guns, knives, explosives, etc) concealed beneath clothing by means of millimeter-wave imaging is becoming a new hotspot in security field. However, the millimeter-wave images are inherently blurred, and some high-frequency components are missed in the process of sampling or scanning. This paper concerns the fusion of millimeter-wave and visible images on the basis of analysis and improvement to the algorithms mentioned in literatures. Moreover, the concealed contraband are identified and located based on the millimeter-wave images and fused images respectively.The principle and shortage of traditional weighted average algorithm is analyzed in this paper, and then a new image fusion algorithm based on enhancement on sensitive areas zone is proposed. In addition, the Contourlet transform is introduced into the fusion of millimeter-wave and visible images, making up the deficiency of incompetence in direction when adopting wavelet transform. The fusion rules for different levels are also discussed, and an improved fusion rule is put forward by analyzing some fusion rules mentioned in literatures. It is indicated that the space quality and information of fused images are improved.The algorithms of identification and localization of concealed contraband based on the fused images are investigated. The moment invariant of concealed contraband is calculated in the millimeter-wave images, and the concealed contraband is identified by the fuzzy C means clustering (FCM) algorithm. Then the temples are extracted from the contraband, and the contraband is located through matching templates in the fused images. Ultimately, the feasibility of identification and location is validated via experiments.