ID |
原文 |
译文 |
23285 |
为了解决 FDK 重建算法在锥束 CT 重建中运算量大,耗时较多,以及针对不同的应用环境提供不同分辨率的 3 维医学图像问题,该文提出一种基于小波变换的多分辨率锥束 CT 图像快速 3 维重建算法。 |
To solve the large amount of computation, time-consuming problems of the FDK reconstruction algorithm for cone beam CT reconstruction, and different resolutions for different application environments of 3D medical image, this paper proposes a fast reconstruction algorithm of multi-resolution cone beam CT image based on wavelet transform. |
23286 |
首先对采集到的投影图像进行相应尺度的小波变换,得到各尺度小波分解系数,选择相应尺度的小波系数进行 FDK 重建,可以得到相应低分辨率的 3 维图像数据,还可根据需要由得到的低分辨率重建数据分别沿着径向取断层图像,进行相应的小波逆变换,进而得到高分辨率的 3 维图像数据。 |
Firstly, the corresponding wavelet transform for projection images are obtained, and the corresponding scale wavelet coefficients are selected for FDK reconstruction. Thus, 3D image data of the low resolution are obtained. According to need, the high resolution 3D image data can also be obtained by the inverse wavelet transform of the radial images obtained from low resolution. |
23287 |
实验数据表明,该方法不仅能够得到不同分辨率的 3 维图像数据,而且相较于传统的 FDK 算法生成分辨率相同、精度相近的高分辨率 3 维图像数据,重建速度可以提高 1 倍以上。 |
The experimental data shows that this method can not only provide a different resolution of the 3D image data, but also increase the reconstruction speed more than one times when the same resolution and similar precision high resolution 3D image data is obtained compared with the traditional FDK algorithm. |
23288 |
该文针对待拼接图像间的亮度差异而导致拼接图像存在拼接缝的问题,提出一种基于 JND(Just Noticeable Difference)模型的新的亮度拼接缝去除方法。 |
An efficient approach based on Just Noticeable Difference (JND) model is proposed to solve the seamline problem caused by illuminate difference. |
23289 |
该图像拼接缝优化算法分为 3 个步骤。 |
The seamline optimization method operates via a three-stage approach. |
23290 |
将待优化图像变换到 HSV 空间,并提取 V 分量; |
In the first stage, the image is transformed into Hue, Saturation, Value (HSV) color space, the V weight is selected for the follow-up stage, and the width of correct region is conformed self-adaptively. |
23291 |
在 V 分量图中自适应确定过渡区域宽度,根据 JND 模型计算区域内各点的优化阈值确定优化量大小并进行优化; |
The second stage is using JND model to calculate the optimize V weight. |
23292 |
将优化后的 V 分量与分离的 H, S 分量结合进行逆变换即可得到最终优化的拼接图像。 |
Last stage is transforming HSV into RGB to get final optimized image. |
23293 |
实验结果表明,该算法能够实现图像亮度的平滑过渡,在亮度差较大的情况下也能得到较好结果。 |
Extensive experimental results using natural images indicate that the proposed method can remove the seamline efficiently. |
23294 |
同时,该算法能够有效地避免常用基于 RGB 空间的拼接缝去除方法所带来的偏色问题,具有更大的适用范围。 |
The effect of seamline removal is better even the brightness difference is great. Meanwhile, the proposed method can avoid the color cast problem caused by RGB color space. |