摘要
由于文字图像的特殊性,产生的特征点数目少,且不同文字产生的特征向量存在强干扰性,导致匹配结果不理想。对SIFT算法进行研究,提出一种对文字模板匹配更有效的SIFT改进算法,改进后的算法能有效增加特征点的数目,消除非匹配点的干扰。经实验测试,SIFT改进算法比原来算法明显提高了文字模版匹配的准确率。
Using the SIFT Algorithm to extract cannot get enough feature points due to the par- ticularity of the text images. And strong inferences between different words lead the bad matc- hing results. This paper made a research on the SIFT algorithm and proposed a more efficient SIFT algorithm for text images template matching. The modified algorithm can effectively in- crease the number of feature points and eliminate the interference of non-matching points. It can effectively improve the accuracy of the text template matching as proved by laboratory test.
出处
《河北农业大学学报》
CAS
CSCD
北大核心
2013年第4期120-123,共4页
Journal of Hebei Agricultural University
基金
河北农业大学理工基金项目(LG20120604)