反馈式ICM神经网络和FPF相结合的剪影识别  

Silhouette recognition using feedback ICM neural network and FPF

作　　者：李迪 徐光柱[1,2] 王亚文 雷帮军[1,2] 邹耀斌[1,2] 夏平[1,2] LI Di;XU Guang-zhu;WANG Ya-wen;LEI Bang-jun;ZOU Yao-bin;XIA Ping(College of Computer and Information Teehnology,China Three Gorges University,Yiehang 443002,China;Hubei Key Laboratory of Intelligent Vision Based Monitoring tor Hydroelectric Engineering,China Three Gorges University,Yichang 443002,China)

机构地区：[1]三峡大学计算机与信息学院,湖北宜昌443002 [2]湖北省水电工程智能视觉监测重点实验室(三峡大学),湖北宜昌443002

出　　处：《广西大学学报（自然科学版）》2018年第3期1089-1096,共8页Journal of Guangxi University（Natural Science Edition）

基　　金：国家自然科学基金资助项目(61402259;U1401252);湖北省自然科学基金创新群体计划项目(2015CFA025)

摘　　要：为能有效地表示目标轮廓,同时在泛化与鉴别之间做好平衡,将具有生物视觉神经系统的交叉视觉皮层模型神经网络(intersecting cortical model nueral network,ICMNN)与分数幂指数滤波器(fractional power filter,FPF)相结合,设计了一种利用反馈式ICMNN分离目标轮廓,并使用FPF实现快速相关的剪影识别系统。首先利用ICMNN对原图像处理得到边缘激发的剪影图像,接着利用经训练集合训练好的FPF对ICMNN输出进行相关运算,得到相关面。为了有效减少相似形状带来的误识别错误,将首次检测到的相关峰值点所对应的原图区域通过反馈方式进行增强,再把增强的结果馈入系统,多次迭代后,可有效减少误识别。研究结果表明:该剪影识别系统能较好地应对目标旋转与尺度变化带来的影响,具有较好的应用价值。In order to describe objects contour validly and consider the trade-off between its generalization and discrimination, by combining the intersecting cortical model neural network （ICMNN） and fractional power filter （ FPF）, a silhouette recognition system has been designed. It uses feedbaek intersecting cortical model neural network to separate target contour and fractional power filter to achieve tast related silhouette recognition. First, ICMNN creates a pulse image containing encouraged object edge. Then, the edge image is correlated with the trained FPF which is created by training images to get a correlation surface. In order to decrease the talse recognition rate caused by different class objects but with similar shape, the correlation image is enhanced at the location of peak point detected first, and the enhanced resuh is feedbaeked to the system. The error recognition can be decreased dramatically alter several iterations. The experimental resuhs show that the proposed silhouette recognition system has good performance for target rotation and sealing, and has good application values.

关 键 词：剪影识别 交叉视觉皮质模型神经网络 反馈式交叉视觉皮质模型 分数幂指数滤波器 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]