主动毫米波成像隐私保护算法

发布时间：2018-06-27 08:00

本文选题：毫米波 + 隐私保护　；参考：《红外与毫米波学报》2017年04期

【摘要】：针对主动毫米波成像安检系统的个人隐私保护问题,提出了一种不受身高、站姿影响的毫米波人体图像隐私保护算法.通过预处理将毫米波图像背景噪声弱化,增加人体目标边缘细节.使用二值化和形态学处理进一步区分背景与人体目标,根据身高划分的人体部位比例与像素点遍历求最值法实现隐私部位的初步定位,可以满足标准站姿下不同身高受检人的隐私部位定位需求.算法所包含的定位复查校准模块通过对隐私位置条件的循环判断与调整,解决了非标准站姿下隐私部位的定位.最后,根据已定位的坐标进行模糊化处理,实现保护受检人隐私的功能.算法已应用于实际产品,与常规隐私保护算法相比,可移植性高,运行速度快,定位准确
[Abstract]:Aiming at the problem of personal privacy protection in active millimeter wave imaging security system, a millimeter wave human body image privacy protection algorithm is proposed, which is independent of height and posture. The background noise of millimeter wave image is weakened by preprocessing, and the edge details of human target are increased. Binarization and morphological processing are used to further distinguish the background from the human object, and the initial location of the private part is realized by the method of maximum value of the human body part proportion and pixel traversal of height division. It can meet the need of privacy location of people with different height under standard standing posture. The location recheck calibration module included by the algorithm solves the location of private position under non-standard standing position through the cyclic judgment and adjustment of the privacy position condition. Finally, the fuzzy processing is carried out according to the position coordinates to realize the function of protecting the privacy of the subjects. The algorithm has been applied to practical products. Compared with the conventional privacy protection algorithm, it has the advantages of high portability, fast running speed and accurate location.
【作者单位】：中国科学院上海微系统与信息技术研究所中科院太赫兹固态技术重点实验室;中国科学院大学;
【基金】：国家自然科学基金(61671439) 中国科学院院地合作项目(YDJDBNJ-2012-027) 上海市自然科学基金(13ZR1447400)~~
【分类号】：TP309;TP391.41

【相似文献】