中文摘要：

基因芯片图像噪声处理决定了后续处理与分析的精度和准确性．高亮斑点噪声是一种亮度很高、成块状的特殊噪声，应用传统噪声处理方法无法滤除．针对高亮噪声的特点，首先利用形态学碟形结构元素进行开运算来评价图像是否含有噪声，然后利用阈值分割去除噪声．提出两种阈值获取方法，一种是基于最大类问方差的自适应获取方法，一种是固定阈值法．通过大量定性和定量实验比较分析，结果表明，两种方法都可以有效除高亮斑点噪声．

英文摘要：

Microarray image noise reduction determines the precision and veracity of microarray data processing and information extraction. Highly fluorescent noise is a type of special noise with higher brightness and shape in block which can＇t be removed by traditional methods. According to the feature of it, this paper proposes that using mathematical morphology opening operator with disk structure element to evaluate image quality, then suppress the noise with thresholding segmentation techniques. Two methods are put forward to get threshold：one is the adaptive method based on maximum variance between clusters, another is the fixed means. Through comparison and analysis, the lots of experiment results show that both two methods can effectively remove the highly fluorescent noise and preserve edge and detailed information.