Body Concealed Carry Knives, Firearms, Hazardous Materials, Thz Image Recognition

Terahertz (THz) imaging is a technology which use THz radiation (refers to the far-infrared electromagnetic radiation which wavelength is in the range of 30μm～3mm) for illumination source. THz radiation can easily penetrate and image inside most dielectric materials, such as clothing, cardboard boxes and other packaging materials. Besides THz radiation has a remarkable advantage for imaging compared with X-ray is safe energy range without ionization. So it will not do any damage to the detected materials or human body. These features make THz imaging play an important role in several areas, such as security-inspection, non-destructive detection for materials.Based on the characteristics of THz imaging , this paper realizes the auto identification to hidden knives and pistols from terahertz image of body by using digital imaging processing technology. The main results obtained are as follows:1. This paper takes pretreatment for terahertz images of body with hidden knives or pistols.(1) Gray-scale transformation. This paper uses Gray-scale transformation to deal with the THz image in order to adapt to different light conditions and elimination the significant changes of gray-scale.(2) Image smoothing. This paper selects the median filtering method with template size is 3x3 to deal with the THz image after compares and analysis the filter effect of two commonly used methods of smoothing(average filtering and median filtering).(3) Image segmentation. Dangerous objects should be separated from the background in order to extract the morphological characteristics of knives and pistols effectively. Images can be viewed as combination of different gray value. Histogram analysis method used to achieve the best image segmentation threshold. It proves that targets can be separated from the background.(4) Edge detection. Edge extracted should be taken for the feature extraction of the pistol. This paper compares the test results of several classical edge detectors. The edge of pistol extracted by LOG operator is good at continuity and integrity.2. This paper analyzes different morphological features of knives and pistols and selects characteristic parameters are the following:(1) Area (A), perimeter (L), length (H), width (W) and rectangular degrees (R) are selected as the main morphological characteristics of knives.(2) Area (A), length (H), width (W) and the angle between barrel and grip (J) are selected as the main morphological characteristics of pistols.3. Identification verification. This paper designs identification programs based on the different characteristic parameters extracted. Experimental results prove that the proposed method can distinguish knives or pistols from lighters, pens, keys, ornaments and other items which often brought with human body effectively. Eventually realizes the auto identification to hidden knives and pistols from terahertz images of human body.The auto identification of hidden knives and pistols from human body's terahertz images by using image recognition technology not only reduces security staff's labor intensity but also improves the efficiency of security checks. More importantly , it can enhances the reliability of safety inspections, discovers the potential security risks in time, protects people's personal and property security effectively and maintains social harmony and stability.