摘要
从理论上研究了图像在发生亮度和对比度变化、镜像或转置变化时,其DCT变换后的矩阵元素之间具有相似性。并在该特性基础上,提出了基于DCT变换的自适应图像匹配算法。鉴于图像匹配标准和抗噪声干扰的鲁棒性,可以只考虑主要能量所存在的区域。另外,图像经过DCT变换后的主要能量集中在左上角,因此,只匹配DCT变换后左上角子矩阵,即可以满足匹配要求。实验证明该方法十分有效。
The similarity of the element of matrix after DCT is studied by theory with changing of brightness or contrasts, and with transformation of mirroring or transposing. On the other hand, by making full use of the characteristic, an adaptive matching approach based on DCT is proposed. With respect to matching criterion and robustness of noise interference, attention is mainly given to a majority of energy, so it can meet the demand when only matching a few elements of matrix on the upper-left corner after transform. Experimental result proves that the method based on DCT is perfect.
出处
《红外与激光工程》
EI
CSCD
北大核心
2004年第6期619-621,641,共4页
Infrared and Laser Engineering
关键词
图像匹配
离散余弦变换
自适应算法
Adaptive algorithms
Cosine transforms
Image analysis
Image quality
Mathematical operators
Matrix algebra
Spurious signal noise