轴向运动目标的热光高阶鬼成像研究

提出一种基于B样条概率密度函数(PDF,probability density function)估计的复杂纹 理图像分类识别方法,主要包括图像纹理PDF特征学习、 表征以及纹理类型识别3个步骤。在图像纹理PDF特征学习和表征中,引入各向异性高斯导 数方向滤波器获得纹理图像多尺度和 多方向空间结构；然后基于预先固定的B样条基函数,将图像空间结构PDF估计转化为与 基函数相对应的权值向量估计；之 后采用最远邻聚类方法,获得图像空间纹理结构的PDF特征字典库；最后采用最近邻方法, 获得各类纹理在特征字典库上的直方图 分布表示。在纹理类型识别阶段,基于直方图距离测量结果实现纹理图像分类识别。在不同 纹理图像数据库上进行了大量的验证性和对比性实验,实验结果表明所提方法的有效性 和优越性。

Research of high-order thermal ghost imaging for an axial moving target

In order to investigate the impact of high-order intensity correlation of light field on axial moving t arget ghost imaging,this paper uses high-order ghost imaging theory in the research of axial moving target ghost imaging.Firstly,we e stablish a mathematical model of ghost imaging for axial moving target based on ideal thema l high-order intensity fluctuation,and the high-order correlation function is obtained.Then the simulation of mathematical model is carried out by setting different parameters,and the sim ulation images are given.Finally,we compare the image quality by using an evalution parameter o f peak signal to noise ratio (PSNR),and then we analyze and discuss th e silulation images based on the evalution method.The results indicate that high-order intensity correlation of the light field can increase the background noise fluctuation with the increase of the reference beam orde r,and the qulity of reconstructed image becomes worse.Instead ,there is little effect on the image quality with the increase of the signal beam order,but the image quality of ghost imagin g for axial moving target is also degraded.In a word,the best quality can be obtained by the lowest -order ghost imaging.