摘要
为了提升特征提取效果,提出了基于K-L变换的无人机巡检图像特征提取方法。利用双边滤波算法去除无人机巡检图像噪声,通过Retinex算法增强去噪后图像亮度,提升图像清晰度。实验证明:该方法可有效去除图像内部噪声,增强图像亮度,提升图像清晰度;可有效提取图像特征,样本维数为3时,特征值与特征向量偏离度最低;特征提取时的真实接受率高至0.9以上。
In order to improve the effect of feature extraction,a feature extraction method of UAV inspection image based on K-L transform is proposed.The bilateral filtering algorithm is used to remove the noise of UAV inspection image.The Retinex algorithm is used to enhance the brightness of the denoised image and improve the image definition.Experiments show that this method can effectively remove the internal noise of the image,enhance the brightness of the image and improve the definition of the image.Besides,this method can effectively extract image features.When the sample dimension is 3,the deviation between feature value and feature vector is the lowest.The real acceptance rate of feature extraction is more than 0.9.
作者
王维佳
白景坡
陶俊
WANG Wei-jia;BAI Jing-po;TAO Jun(Anhui Jiyuan Software Co.,Ltd.,Hefei 230038,China;State Grid Information and Communication Group Co.,Ltd.,Beijing 102209,China)
出处
《信息技术》
2023年第4期35-40,共6页
Information Technology
基金
国网信息通信产业集团有限公司项目(5268002XXX2W)。
