基于BCS-SPL压缩感知算法的纸病图像重构

Paper Disease Image Reconstruction Based on BCS-SPL Algorithm

关键词:  压缩感知  BCS-SPL重构算法  纸病图像重构

Key Words:compressed sensing  reconstruction algorithm of BCS-SPL  reconstruction of paper disease image

摘要:随着造纸工业纸机速度和纸幅宽度的增长,传统的纸病检测处理方式面临着图像数据传输量剧增,纸病检测系统难以实现实时性处理的问题。压缩感知理论能够有效降低数据的采样量,但将压缩感知应用于二维纸病图像时,面临着重构纸病图像质量不高的问题。本研究采用分块压缩感知（BCS）-平滑投影Landweber（SPL）重构算法对纸病图像进行重构,并着重研究了该算法在不同采样率和不同图像分块大小下的重构效果。实验结果表明,在压缩感知框架下,通过BCS-SPL算法重构的低采样率纸病图像具有较高的图像质量,有效降低了纸病图像数据的传输量。

Abstract:With the growing of the speed and the width of paper machine, the traditional paper disease detection faces the problem of transfering a large number of data and the real-time processing. Compressed sensing theory can effectively reduce the amount of data, but the quality of reconstructed paper disease image is not good when it is applied to two-dimensional paper disease image. In this paper, we used the BCS-SPL reconstruction algorithm to reconstruct the paper disease image, focusing on the reconstruction result of the algorithm under different sampling rates and different block sizes. The experimental results showed that in the compressed sensing framework, the low sampling rate paper disease image reconstructed by BCS-SPL algorithm had high image quality, which could effectively reduce the transmission of paper image data．