基于Faster R⁃CNN的无人超市商品自动化识别技术  

Automatic recognition technology of unmanned supermarket goods based on Faster R⁃CNN

作　　者：陆青梅[1] LU Qingmei(College of Innovation and Entrepreneurship,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学创新创业学院,山西太原030051

出　　处：《现代电子技术》2024年第12期145-149,共5页Modern Electronics Technique

基　　金：山西省教育科学“十四五”规划项目:计算能力培养模式研究(GH-220360)。

摘　　要：无人超市商品自动化识别过程中易受到背景复杂化、亮度不均匀、角度多变等的干扰。为此,提出一种基于Faster R-CNN的无人超市商品自动化识别方法。首先利用Haar小波提升模型,将商品图像分为低频图像和高频图像;然后通过仿生彩色图像法对图像进行增强处理,并采用Faster R-CNN中的特征融合结构,将图像深度信息与浅度信息融合到一起;最后将融合的特征输入到自动化识别网络中,输出自动化识别结果。实验结果表明,所提方法的识别效率高、图像增强效果好、抗噪能力强。In the process of automatic identification of goods in unmanned supermarket,it is easy to be disturbed by the problems of complex background,uneven brightness,veriable angle and so on.Therefore,a method of unmanned supermarket goods automatic commodity identification based on faster R-CNN is proposed.The Haar wavelet lifting model is used to divide the commodity image into low-frequency image and high-frequency image,and then the bionic color image method is used to enhance the image.The feature fusion structure in faster R-CNN is used to intergate the image depth information and shallow information.The fused features are input into the automatic recognition network and the automatic recognition results are output.The experimental results show that the proposed method has high recognition efficiency,good image enhancement effect and strong anti-noise ability.

关 键 词：Faster R-CNN 无人超市 自动化识别 HAAR小波 商品图像 图像增强 中值滤波 小波系数 

分 类 号：TN911.73-34[电子电信—通信与信息系统] TP753[电子电信—信息与通信工程]