摘要
针对低照度图像环境下目标检测精度不高这一问题,该文提出了基于改进SSD(single shot multi-box detector)的低照度图像目标检测算法。首先基于Retinex理论的图像增强算法进行原始低照度图像的增强;然后在SSD模型基础上,联合利用原始图像与增强图像之间特征差异性,设计了双分支SSD结构并采用ResNet50网络替换原始VGG16特征提取网络;最后在双分支结构中嵌入一种差分特征融合模块(DFF),使模型对互补特征有更好的提取效果以此来提高算法对低照度图像目标检测精度。实验结果表明,该文所提出的方法在低照度图像环境下,与主流的检测算法相比,检测精度能达到82.39%。
Aiming at the problem of low target detection accuracy in low illumination environment,a low illumination target detection algorithm based on modified SSD(single shot multi box detector)is proposed in this paper.Firstly,the image enhancement algorithm based on Retinex theory enhances the original low illumination image;Then,based on the SSD model,using the feature difference between the original image and the enhanced image,a double branch SSD structure is designed,and the ResNet50 network is used to replace the original VGG16 feature extraction network;Finally,a differential feature fusion module(DFF)is embedded in the double branch structure to make the model have better extraction effect on complementary features,so as to improve the accuracy of the algorithm for low illumination target detection.The experimental results show that the detection accuracy of the proposed method can reach 82.39%compared with the mainstream detection algorithm in low illumination environment.
作者
吕东
张鑫港
LV Dong;ZHANG Xin-gang(School of Information Science and Technology,Qingdao University of Science and Technology,Qingdao 266061,China;Faculty of Geomatics,Lanzhou Jiaotong University,Lanzhou 730070,China)
出处
《自动化与仪表》
2022年第5期53-58,69,共7页
Automation & Instrumentation
