Abstract: In order to improve visual field of photos
an improved and automatic image mosaic algorithm is presented based frequency and time domain features. Phase correlation is used to sort the unordered image sequence and estimate the overlapping area
in which improved Harris operator is used to extract corners. Then
the initial feature point pairs are obtained with bidirectional greatest correlative coefficient
and the false feature point pairs are rejected by RANSAC. Finally
image mean combined linear weight function is used to implemented image mosaic. Experiment results show that this algorithm exceeds existing ones at stability and matching speed about 40%.It can efficiently solve the difficulty in confirming corresponding points and gain satisfying visual effect in image mosaics with notable illumination difference.