A Design Scheme of Multispectral Criminal Detector

A new design scheme of multispectral criminal detector is proposed to overcome the problem that the existing hyperspectral criminal detection instrument possess tedious extraction of physical evidence information, the expensive equipment and so on. First, CIS(Contact Image Sensor) is selected to emit five sources of red, green, blue, infrared and ultraviolet, and so the image information produced by light source is obtained. Then, the high precision image information and high-speed data transmission are acquired by FPGA. Finally, the Gabor filter algorithm is used to extract feature information and support vector machine training for realizing the extraction of fully automated material evidence. Experimental results show that the output resolution of the proposed design scheme is as high as 1200x2400DPI, and material evidence extracting takes 12.5 seconds, furthermore, the accuracy rate of automatic material evidence recognition is as high as 92%. The proposed design scheme can quickly and accurately realize the fingerprint, alter or cover the handwriting automatically.