文摘
|
样点自动识别是生物芯片信息自动提取的关键。根据样点、噪声和背景特征的关系提出一种新的自动识别方法。使用数学形态学和均值算子相结合的方法实现图像的滤波增强和背景亮度的估计;通过对功率谱的分析实现图像的倾斜校正和样点中心的网格定位;利用样点边缘亮度与均方差的特征实现样点中心和半径的校正。多幅生物芯片处理的实验证明该方法具有良好的抗噪声能力和弱信号辨识能力,能快速、准确地实现样点自动识别. |
其他语种文摘
|
Automatic-identification of spots is the key for automatic-extracting information from biochip images. A new approach for automatic-identification is proposed based on the relations of spots, noises and backgrounds. Morphological and mean filtering was used to filter and enhance images and estimate intensity of background. Spectrum of power was used to correct tilting and perform spots gridding automatically. Considering intensity and standard deviation of edge, a new technique to correct center and radius was introduced. The approach has been tested in a variety of biochip images, and better resistant performance of noises and identification of weak signals are realized. It can automatic-identify spots quickly and accurately. |
来源
|
光电工程
,2006,33(3):95-100 【核心库】
|
关键词
|
生物芯片
;
图像处理
;
数学形态学
;
功率谱
;
边缘分析
|
地址
|
中国科学院光电技术研究所, 四川, 成都, 610209
|
语种
|
中文 |
文献类型
|
研究性论文 |
ISSN
|
1003-501X |
学科
|
自动化技术、计算机技术 |
基金
|
中国科学院西部之光人才培养计划
|
文献收藏号
|
CSCD:2340057
|