实时VR场景下多尺度数字全息成像仿真
Simulation of Multi-Scale Digital Holography in VR Scene
辜昕宇 1黎鹏2
作者信息
- 1. 江西师范大学科学技术学院,江西 南昌 332020
- 2. 南昌大学,江西 南昌 332020
- 折叠
摘要
在数字全息成像的过程中,受光电传感器件结构参量、光照强度以及传输方向等因素的影响,导致全息图成像的亮度、分辨率降低.为了提高多尺度数字全息成像的真实度,提出了实时VR场景下多尺度数字全息成像方法.通过定义图像像素点分布的概率函数,计算多尺度数字图像像素的显著性,根据图像像素特征的分布方差,提取出多尺度数字图像像素点.利用小波变换系数识别多尺度数字图像的噪声,基于图像噪声信号的方差分布,精细化处理多尺度数字图像,结合小波反变换,构建多尺度数字图像去噪模型,得到去噪后的图像.根据多尺度数字全息成像模型,得到图像增强的梯度场,在实时VR场景下,引入加权融合的方法,得到图像的高频细节分量,利用图像融合规则,生成多尺度数字图像.实验结果表明,所提方法能够生成VR图像,并将成像真实度和清晰度分别提高到 90%以上和 92%以上.
Abstract
Due to the influence of the structural parameters of photoelectric sensors,light intensity and transmis-sion direction,the brightness and resolution of the hologram imaging may be reduced in the process of digital holo-graphic imaging.In order to improve the trueness of multi-scale digital holographic imaging,a new method in real-time VR scenes was proposed.Firstly,we calculated the significance of multi-scale digital image pixels by defining the probability function of pixel distribution.According to the distribution variance of pixel characteristics,we extrac-ted the pixels of multi-scale digital images.And then,we used the wavelet transform coefficients to identify the noise of multi-scale digital images.Based on the variance distribution of noise signals,we refined multi-scale digital ima-ges.Moreover,we built a denoising model based on the wavelet inverse transform,thus obtaining the denoised images.According to the multi-scale digital holographic imaging model,we could obtain the gradient field of image enhance-ment.In the real-time VR scene,we used the weighted fusion method to obtain the high-frequency detail components.Finally,based on the fusion rule,multi-scale digital images were generated.Experimental results show that the proposed method can generate VR images,and improve the trueness and clarity of images to more than 90%and 92%respectively.
关键词
实时虚拟现实场景/全息成像/像素点提取/去噪模型/多尺度/真实度Key words
Real-time VR scene/Holographic imaging/Pixel extraction/Denoising model/Multiscale/Trueness引用本文复制引用
基金项目
江西省高等学校人文社会科学课题(2021)(YS21210)
出版年
2024
NETL
NSTL
万方数据