Abstract
The correlation of edge gradient was adopted to replace the common gray correlation as the target function of image co-registration. The contour pixels of two images were described as two rows of 2D-sequence which had time-delay relationship through frame contrasting and then the optimal transform of the two images was obtained from the corresponding points in the sequences through the least squares method. The co-registration result of the magnetic resonance-positron emission tomography (MR-PET) images of cranium brain showed that the accuracy of using the edge gradient correlation as target function is better than that of using gray correlation. And the method that contrasts the pixels in the contour to get the optimal transform relationship is less time consumed than other methods, such as genetic algorithm and searching the whole data space, while the accuracy is also guaranteed.
Original language | English (US) |
---|---|
Pages (from-to) | 627-631 |
Number of pages | 5 |
Journal | Zhejiang Daxue Xuebao (Gongxue Ban)/Journal of Zhejiang University (Engineering Science) |
Volume | 42 |
Issue number | 4 |
State | Published - Apr 2008 |
Keywords
- Correlation function
- Edge detection
- Image co-registration