基于分水岭分割的超光谱图像虚拟重建方法  

Virtual Reconstruction of Hyperspectral Image Based on Watershed Segmentation

作　　者：于翠萍[1] 赵志刚 YU Cui-ping;ZHAO Zhi-gang(Eastern Liaoning University,Dandong Liaoning 118001,China;Qilu University of Technology(Shandong Academy of Sciences),Ji'nan Shandong 250014,China)

机构地区：[1]辽东学院,辽宁丹东118001 [2]齐鲁工业大学(山东省科学院),山东济南250014

出　　处：《计算机仿真》2021年第8期386-390,共5页Computer Simulation

摘　　要：针对超光谱图像需要较大存储空间,会产生过度分割的现象,提出了一种基于分水岭分割的超光谱图像虚拟重建方法。通过对超光谱图像的分析,以VisualC++作为虚拟设计平台,给出虚拟重建整体框架,选用FSL作为总线接口来完成设备之间的通信传输。为提高超光谱图像的传输速度,减小图像储存空间,采用普间DPCM和整数小波变换对图像压缩,同时使用分水岭变换法分割处理压缩后的微弱边缘图像,避免过度分割。以视觉传达原则为基础,采用空间指示函数完成超光谱图像虚拟重建。实验结果表明,虚拟重建结果稳定性更强,重建速度更快,精准度更高,更具实际应用价值。A virtual reconstruction method of hyperspectral image based on watershed segmentation is proposed in this work for reducing the over segmentation of hyperspectral image due to large storage space.Firstly,hyperspectral images were analyzed,Visual C++was set as the virtual design platform to give the overall framework of virtual reconstruction,and FSL was adopted as the bus interface to complete the communication between devices.Secondly,universal DPCM and integer wavelet transform were introduced to decompress the image in order to improve the transmission speed of hyperspectral image and reduce the image storage space.Then,watershed transform was used to segment the compressed weak edge image for avoiding over segmentation.Finally,according to the principle of visual communication,spatial indicator function was utilized to achieve the virtual reconstruction of hyperspectral image.Experimental results show that the method has stable virtual reconstruction results,fast reconstruction speed,high accuracy,and excellent practical application value.

关 键 词：超光谱图像 整数小波变换 分水岭分割 视觉传达 

分 类 号：TN304.2[电子电信—物理电子学]