倾斜校正是全自动生物芯片图像处理必不可少的环节。针对以往倾斜校正算法耗时过长,本文提出一种快速倾斜校正算法。算法首先用细胞神经网络(CNN)寻求各个样点的中心,然后进行Hough变换,再计算方差来寻求倾斜角,最后利用CNN灰度图像旋转模板进行倾斜校正。本算法利用了细胞神经网络并行处理的特性,并充分考虑了生物芯片图像的特点。理论分析与试验结果显示本文算法能够准确高速地完成倾斜校正。

Skew correction algorithm is indispensable in fully automated microarray image processing. Since most previous image skew correction algorithms for microarray were time-consuming,a fast skew correction algorithm was proposed based on Cellular Neural Networks(CNN). The first step was to obtain the center points of all spots by applying a series of CNN templates. The second step was to find out the skew angle by performing Hough transform and calculating deviation. The final step was to correct skew by CNN grey-scale image rotation template. Our algorithm utilized the parallel processing attributes of CNN and the characteristics of microarray image. Experiment results and theoretical analysis show that the algorithm works accurately and fast.