基于阈值分割和边缘检测的枪支TH_z图像识别

Feature Recognition of Pistol in Terahertz Image Based on Threshold Segmentation and Edge Detection

针对太赫兹(THz)图像的特点,采用阈值分割和边缘检测等数字图像处理技术将枪支分割出来,然后提取枪支的有效特征,并对人体携带枪支的THz图像进行自动识别和判断。若发现有危险品或可疑物品,则发出警报信号。实验结果表明,该方法能够有效地将人体隐蔽携带的枪支类危险品同钢笔、钥匙、挂饰等人体通常携带的其它物品区别开来,实现了安全检查中对枪支武器类违禁品THz图像的自动识别。

According to the characteristics of THz images, some digital image processing methods such as threshold division and edge detection are used to segment the pistol from a THz image. Then, the effective signatures of the pistol are extracted and automatic recognition and judgment are implemented on the image. If hazardous or suspicious goods are found, an alarm signal is generated. The experimental result shows that this method can effectively discriminate the pistols concealed in clothes from other metal goods such as pens, keys and ornaments which people usually carry. It can be used to automatically identify the contraband goods such as pistols in the THz images in a security check.