基因芯片荧光靶点图像检测识别算法改进研究

在基因芯片分析系统中,基因芯片荧光靶点图像的正确检测识别是基因特异性表达信息提取的必要前提。在荧光靶点检测识别过程中,由于沾污、瑕疵、离焦等因素的影响,荧光靶点图像的信噪比很低,很容易将污点、基片瑕疵等噪声点误识别为荧光靶点,而将沾污的荧光靶点误识别为噪声点。在原算法基础上,为进一步降低误识别率和提高检测精度,提出基于靶点分割图像重心和目标背景面积比的改进的荧光靶点检测识别算法。实验结果表明,与原算法相比,采用新算法将正确识别率提高到90%以上。

In gene chip(i.e.,Micro-arrays) analysis system,the accurate detection and recognition of arrayed fluorescent spots images is the prerequisite step for gene differential expression information distilling.In the process of spot detection and recognition,the spot image signal noise ratio degradation caused by the effects of contamination,flaws and defocus directly leads to the disappointing results that the blurs and flaws are recognized as true spots but blurred spots as noise dots.To further lower the erron...