基于CNN算法在纸病图像增强中的应用  

Application of CNN Algorithm in Paper Disease Image Enhancement

作　　者：彭丽娟[1] 王军红[1] 王金炜[1] PENG Lijuan;WANG Junhong;WANG Jinwei(Shaanxi Institute of International Trade and Commerce,Xi’an 712000,China)

机构地区：[1]陕西国际商贸学院,陕西西安712000

出　　处：《造纸科学与技术》2023年第2期75-77,共3页Paper Science & Technology

摘　　要：为识别低照度条件下纸病图像的纸病区域和背景区域,纸病的存在严重影响到纸张的质量和纸质产品生产的效率,因此提出一种CNN算法优化的纸病图像增强方法。首先对造纸厂生产线上工业相机采集的纸张图像进行预处理,然后利用CNN算法提取视觉图像的特征,重点设计了基于CNN的缺陷检测分类方案、CNN网络计算过程和CNN网络训练。经验证,算法的对比度增强效果为所有对比算法中最优,完全满足纸病图像增强的质量要求。In order to identify the paper disease area and background area of the paper disease image under the condition of low illumination,the existence of paper disease seriously affects the quality of paper and the efficiency of paper product production,so a paper disease image enhancement method optimized by CNN algorithm was proposed.Firstly,the paper images collected by industrial cameras on the production line of paper factory are preprocessed,and then the CNN algorithm is used to extract the features of visual images.The defect detection classification scheme based on CNN,CNN network calculation process and CNN network training are designed emphatically.It is verified that the contrast enhancement effect of the proposed algorithm is the best among all the comparison algorithms,and it fully meets the quality requirements of paper disease image enhancement.

关 键 词：CNN算法 纸病图像 图像处理 样本训练 

分 类 号：TS71[轻工技术与工程—制浆造纸工程]