Analysis method of noise power spectrum for medical monochrome liquid crystal displays
Analysis method of noise power spectrum for medical monochrome liquid crystal displays
复制标题
DOI：
10.1007/s12194-008-0029-y
复制
发表时间：
2008-07-01
影响因子：
1.6
通讯作者：
Matsubara, Kosuke
中科院分区：
文献类型：
--
作者：
Ichikawa, Katsuhiro;Kodera, Yoshie;Matsubara, Kosuke
研究了医用液晶显示器噪声功率谱（NPS）测量的分析方法。利用配备近距离镜头的高性能数码相机对显示在液晶显示器上的均匀图像进行成像，然后通过多种分析方法对图像数据进行nps计算。在一种使用256 x 256像素数据段的二维快速傅里叶变换（FFT）方法（基本方法）中，我们检查了背景趋势校正（BTC）和用于减少傅里叶分析中光谱估计误差的汉宁窗口处理的有效性。为了提高基本方法的频率分辨率，我们研究了两种二维FFT方法，分别使用512 × 512和1024 × 1024像素段。此外，我们还研究了一种具有1024点一维噪声分布的一维FFT方法（1D方法）。在这三种方法中，常用的是二阶多项式拟合的BTC和汉宁窗。使用300万像素（MP）和5万像素单色LCD来评估各自的方法。此外，还将配备新型抗反射表面涂层面板的5万像素LCD原型与传统的5万像素LCD进行了比较。汉宁加窗处理是避免液晶显示器像素结构引起的光谱泄漏误差必不可少的方法。采用1024 × 1024像素段的二维FFT方法和一维方法获得了足够的频率分辨率。提供最可靠的nps的方法是一维法，该方法成功地获得了BTC。
We investigated methods of analyzing the noise power spectrum (NPS) measurement for medical liquid crystal displays (LCDs). Uniform images displayed on the LCDs were imaged with a high-performance digital camera equipped with a close-up lens, and then the NPSs were calculated from the image data by means of several analysis methods. In a method using the 2D fast Fourier transform (FFT) with a 256 x 256 pixels data segment (basic method), we examined the efficacy of a background trend correction (BTC) and a Hanning windowing process used for reducing the spectral estimation errors in the Fourier analysis. To improve the frequency resolution of the basic method, we examined two 2D FFT methods by using 512 x 512 and 1024 x 1024 pixel segments. In addition, we studied a 1D FFT method with 1024-point 1D noise profiles (1D method). In these three methods, the BTC by a second-order polynomial fit and Hanning windowing were commonly applied. A 3-mega-pixel ( MP) and a 5-MP monochrome LCD were employed for evaluating the respective methods. Also, a prototype 5-MP LCD equipped with a new anti-reflection surface-coated panel was compared with the conventional 5-MP LCD. The Hanning windowing process was indispensable for avoiding the spectral leakage errors caused by the pixel structures of the LCD. Sufficient frequency resolution was obtained by the 2D FFT method with the 1024 x 1024 pixels segments and the 1D method. The method which provided the most reliable NPSs was the 1D method, with which the BTC was achieved successfully.