融合视觉传达的模糊激光图像纹理特征提取技术  

Fuzzy laser image texture feature extraction technology for fusion visual communication

作　　者：徐璐[1] 武文英 黄泽军[1] XU Lu;WU Wenying;HUANG Zejun(Taiyuan University,Taiyuan 030032,China;Institute of resources and environment,Shanxi Agricultural University,Jinzhong Shanxi 030801,China)

机构地区：[1]太原学院,太原030032 [2]山西农业大学资源环境学院,山西晋中030801

出　　处：《激光杂志》2025年第4期109-114,共6页Laser Journal

基　　金：山西省教育科学“十四五”规划2022年度课题(No.GH-220036)。

摘　　要：由于散射、衍射等物理效应或成像设备限制,模糊激光图像的纹理边界不清晰,且其纹理具有多种尺度和方向性,全面且有效地提取纹理特征变得尤为困难。通过视觉传达融合能够结合不同视觉特征的优势,从而更全面地描述模糊激光图像的纹理信息,基于此提出了融合视觉传达的模糊激光图像纹理特征提取方法。将RGB空间中的模糊激光图像转换到利于人眼感知的CIE Lab颜色空间,以增强图像的颜色信息,并通过MSRCR算法和CLAHE算法在保持图像颜色的同时,增强模糊激光图像的对比度;在此基础上,利用Gabor滤波器的复数性质,在多个尺度和方向上提取模糊激光图像的纹理特征信息。实验结果表明,该方法的图像预处理效果好,且特征平均吻合度在0.98以上,结构相似性指数平均值达到了0.84,峰值信噪比平均值达到了27.70 dB,具有较高的纹理特征提取的性能。Due to physical effects such as scattering and diffraction,or limitations of imaging equipment,the texture boundaries of blurred laser images are unclear,and their textures have multiple scales and directionality,making it particularly difficult to comprehensively and effectively extract texture features.Visual communication fusion can combine the advantages of different visual features to more comprehensively describe the texture information of fuzzy laser images.Based on this,a method for extracting texture features of fuzzy laser images by integrating visual communication is proposed.Convert the blurred laser image in RGB space to a CIE Lab color space that is conducive to human perception,to enhance the color information of the image,and use the MSRCR algorithm and CLAHE algorithm to enhance the contrast of the blurred laser image while maintaining the image color;On this basis,utilizing the complex properties of Gabor filters,texture feature information of blurred laser images is extracted at multiple scales and directions.The experimental results show that the image preprocessing effect of this method is good,and the average feature fit is above 0.98.The average structural similarity index reaches 0.84,and the average peak signal-to-noise ratio reaches 27.70 dB.It has a high performance in texture feature extraction.

关 键 词：视觉传达 模糊激光图像 MSRCR算法 GABOR滤波器 纹理特征提取 

分 类 号：TN911.73[电子电信—通信与信息系统]