摘要
为提高安全标识施工指示牌在受到光线、污染等情况下的自动识别效果,研究基于人工智能技术的安全标识施工指示牌自动识别方法。首先采集安全标识施工指示图像,对其进行亮度增强、恢复图像颜色,然后通过小波变换提取图像特征,将其作为卷积神经网络的输入,通过网络的学习和训练,建立指示牌分类识别器,测试结果表明,通过该方法增强的施工指示牌图像纹理清晰、色调均匀,可有效识别各类指示牌,在分辨率以及噪声增加情况下,识别准确率的变化较小,始终在97%以上。
In order to improve the automatic recognition effect of safety sign construction sign under the condition of light and pollution,the automatic recognition method of safety sign construction sign based on artificial intelligence technology was studied.Firstly,the construction indication image safety sign is collected to enhance the brightness and restore the image color.Then,the image features are extracted by wavelet transform and used as the input of convolution neural network.Through the learning and training of the network,the indicator classification recognizer is established.The test results show that the texture of the construction indication image enhanced by this method is clear and the tone is uniform,It can effectively identify all kinds of signs.With the increase of resolution and noise,the change of recognition accuracy is small,which is always more than 97%.
作者
李菊芬
范晶荣
赵明黎
邱薇
徐荣荣
LI Jufen;FAN Jingrong;ZHAO Mingli;QIU Wei;XU Rongrong(Yunnan Transportation Research Institute Co.,Ltd.,Kunming 650000,China)
出处
《自动化与仪器仪表》
2022年第1期185-188,共4页
Automation & Instrumentation
基金
云南省道路运输从业人员管理系统研究与推广应用,编号:180103001。
关键词
人工智能技术
安全标识
施工指示牌
特征提取
分类识别
artificial intelligence technology
safety identification
construction signs
feature extraction
image bright-ness
classification
作者简介
李菊芬(1978-),女,汉族,云南大理,副高级工程师,大学本科,研究方向为交通工程。
