摘要
传统的眼底图像拼合主要采用SIFT算法实现,算法计算量大,效率低,准确率也不高。Harris提取的是图像的角点信息,能较好准确地反应图像的特征信息。作者采用Harris和SIFT算法相结合的方式,采用Harris算法提取的角点信息,并用这些角点提取SIFT特征描述符,且在图像拼合的过程中对算法做出相应的改进,克服了多模式图像中局部梯度信息不一致带来的误差,计算效率高,且拼合效果较好。
On the traditional way using the SIFT algorithm,the computation is heavy while with bad accuracy.Harris algorithm is used to figure out the Corners,which can typically represent the figure＇s characteristic.The improved Harris is combined with SIFT algorithm,the Corners information is figured out through Harris,and the took out the SIFT corresponding descriptors,while overcoming the errors made by the local gradient inconsistence of multimodal image,and achieved well mosaic results at low computation.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第9期3507-3511,共5页
Computer Engineering and Design
