箱包检测可疑区域的自动识别  

Automatically Discerning Interesting Regions in Luggage Inspection System

作　　者：洪金希[1] 吴志芳[1] 刘锡明[1] 刘以思[1] 

机构地区：[1]清华大学核能与新能源技术研究院,北京100084

出　　处：《原子能科学技术》2006年第5期625-628,共4页Atomic Energy Science and Technology

摘　　要：基于箱包中典型物品与可疑物DR图像的灰度差异、典型物品区与可疑区间的灰度方差差异等判断依据,研究建立箱包检测DR图像中可疑区域的自动算法流程。该流程涉及DR图像的二值化,噪声干扰消除,图像形态学腐蚀运算和膨胀运算,可疑区域边界提取,可疑区标定。采用所研制的自动识别算法流程对典型DR图像进行处理。处理后的DR图像中的可疑物被清晰显示和标出。标示出的可疑物的图像边界与实物略有差异,这是本算法流程需进一步完善和改进之处。In the DR luggage image, based on the difference of gray between typical objects and interesting objects, and on the difference of gray deviation between typical regions and interesting regions, an automatically discerning algorithm flow of interesting regions in the DR luggage image was researched and constructed. This flow concerns in the two-valuation of DR image, filtering of noise interference, dilation operation and erosion operation of image morphology, extraction of borders of interesting regions, and demarcation of interesting regions. Processed with the automatically discerning algorithm flow, interesting objects in the DR image can be showed and demarcated clearly. There is also a little of difference between the border demarcated of interesting objects and the border of practical objects, which is the one point still required to be improved and ameliorated in this algorithm flow.

关 键 词：箱包检测系统 自动识别 可疑区域 

分 类 号：TL751[核科学技术—辐射防护及环境保护]