摘要
为自动识别激光图像中的噪声,增强图像处理自动化和智能化水平,提出基于人工智能技术的激光图像中噪声自动识别方法。分析激光图像中噪声类型,构建激光图像噪声自动识别样本数据集;提取小波高频系数直方图不显著系数能量比、噪声能量熵分布梯度特征平面、噪声曲面边界法矢、曲率变化率以及曲面能量离散率五个特征值,作为激光图像噪声特征向量;构建基于最小二乘支持向量机(LSSVM)的激光图像噪声自动识别模型,并采用萤火虫算法优化模型中的关键参数;利用构建的样本数据集对优化后的LSSVM实施训练,将提取出的激光图像噪声特征向量输入训练好的模型中,模型输出结果就是激光图像中噪声识别结果。实验表明,该方法可以精准识别激光图像中的噪声,在识别灵敏度方法具有较好的表现,灵敏度系数可达0.92以上。
To automatically identify noise in laser images and enhance the automation and intelligence level of image processing,an artificial intelligence based method for automatic recognition of noise in laser images is proposed.Analyze the types of noise in laser images and construct a sample dataset for automatic recognition of laser image noise;extract five feature values from the wavelet high-frequency coefficient histogram,including the insignificant coefficient energy ratio,noise energy entropy distribution gradient feature plane,noise surface boundary normal vector,curvature change rate,and surface energy dispersion rate,as laser image noise feature vectors,construct a laser image noise automatic recognition model based on least squares support vector machine(LSSVM),and optimize the key parameters in the model using firefly algorithm,using the constructed sample dataset to train the optimized LSSVM,the extracted laser image noise feature vectors are input into the trained model,and the model output is the noise recognition result in the laser image.The experiment shows that this method can accurately identify noise in laser images,and has good performance in recognition sensitivity methods,with a sensitivity coefficient of over 0.92.
作者
张志东
任瑞仙
段海英
ZHANG Zhidong;REN Ruixian;DUAN Haiying(Shanxi Institute of Technology,Yangquan Shanxi 045000,China)
出处
《激光杂志》
北大核心
2025年第10期162-168,共7页
Laser Journal
基金
山西省自然科学基金项目(No.202303021211153)。
关键词
人工智能
激光图像
噪声识别
特征值
最小二乘支持向量机
萤火虫算法
artificial intelligence
laser image
noise recognition
eigenvalue
least squares support vector machine
firefly algorithm
