基于多方向阈值的超分辨率图像噪声识别仿真  

Simulation of Super-Resolution Image Noise Recognition Based on Multi-Directional Threshol

作　　者：王旭阳[1] 刘世健 WANG Xu-yang;LIU Shi-jian(College of Computer and Communication,Lanzhou University of Technology,Lanzhou Gansu 730050,China)

机构地区：[1]兰州理工大学计算机与通信学院,甘肃兰州730050

出　　处：《计算机仿真》2021年第12期132-135,181,共5页Computer Simulation

基　　金：甘肃省重点研发计划:无人机关键技术研究(18YF1GA060)。

摘　　要：针对当前超分辨率图像噪声识别方法,未考虑获取多方向阈值分割图像,导致超分辨率图像噪声识别时间长、识别精度和识别覆盖率低的问题,提出基于多方向阈值的超分辨率图像噪声识别方法。依据一维函数灰度曲线获取局部阈值,利用灰度波动局部阈值分割法,分割局部阈值图像,分析图像中噪声曲面的曲率变化率、弹性变化率、边界法矢、曲面离散率、外载荷修正量等特征值,将分析出的特征性质整合成一组特征向量,通过分类器进行计算,实现超分辨率图像噪声识别。实验结果表明,所提方法的超分辨率图像噪声识别精度较高,能够有效缩短超分辨率图像噪声识别时间,提高识别覆盖率。The lack of multi-directional threshold segmentation image in super-resolution image noiserecognition method results in long recognition time, low recognition accuracy and recognition coverage. Therefore,this paper puts forward a method of super-resolution image noise recognition based on multi-directional threshold.Based on the gray curve of one-dimensional function, the local threshold was obtained. The local threshold segmenta-tion method of gray fluctuation was applied to segment the local threshold image. The curvature change rate, elasticchange rate, boundary normal vector, surface dispersion rate and external load correction of the noise surface in theimage were analyzed in detail, respectively. The analysis results were integrated into a set of eigenvectors. Accordingto the classifier, these feature vectors were calculated to realize the noise recognition of super-resolution image. Theexperimental results show that the noise recognition of super-resolution image based on this method has high accura-cy, coverage and short recognition time.

关 键 词：多方向阈值 超分辨率图像 噪声识别 图像分割 特征值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]