共聚焦激光检眼镜下眼底图像的非特征配准约束模型研究

A STUDY ON THE CHOOSING OF UNDERLYING MOTION MODEL IN REGISTERING THE FUNDUS IMAGE UNDER CONFOCAL LASER OPHTHALMOSCOPE

研究共聚焦激光检眼镜下不基于特征提取的眼底图像自动配准方法中的运动约束模型,从成像机理上分析共聚焦激光检眼镜下图像对间的运动模式,并分析比较多种实际全局运动模型约束下的配准精度和效率,进而给出一种由粗到细的复合约束模型对眼底图像进行配准。实验结果证实了该模型效果良好。

The underlying motion model between the image sequences was studied in fundus angiography under confocal laser ophthalmoscope. Furthermore, based on theoretical analysis of the inher-ent mechanism of the angiography procedure and comparison of both accuracy and effectiveness of image registration algorithm under several global motion models, a combined motion model strategy with the model parameters dimension increasing from stage to stage is presented. The experimental results of image registration show satisfactory matching, both visually and in sum of squared difference minimization sense.