摘要
复杂光照条件下的光源变化、阴影干扰和反射效应严重影响机器视觉目标识别的准确性。为解决这一问题,研究提出了一种结合图像预处理与深度学习的目标识别方法。将光照自适应图像增强技术与多尺度卷积神经网络(Multi-Scale Convolutional Neural Network,MSCNN)相结合,能显著提高不同光照条件下的目标识别精度。
The changes in light sources,shadow interference,and reflection effects under complex lighting conditions seriously affect the accuracy of machine vision object recognition.To address this issue,the study proposes a target recognition method that combines image preprocessing and deep learning.Combining lighting adaptive image enhancement technology with multi-scale convolutional neural network(MSCNN)can significantly improve target recognition accuracy under different lighting conditions.
作者
李炜深
劳宇晴
邹国涛
LI Weishen;LAO Yuqing;ZOU Guotao(School of Transportation and Economic Management,Guangdong Communication Polytechnic,Guangzhou Guangdong 510800,China;School of Rail Transit,Guangdong Communication Polytechnic,Guangzhou Guangdong 510650,China)
出处
《信息与电脑》
2025年第16期25-27,共3页
Information & Computer
基金
广东大学生科技创新培育专项资金资助项目(项目编号:pdjh2025bc333)。
关键词
机器视觉
目标识别
复杂光照
machine vision
object recognition
complex lighting
