A Partial Intensity Invariant Feature Descriptor for Multimodal Retinal Image Registration
read more
Citations
Retinal Imaging and Image Analysis
Image Matching from Handcrafted to Deep Features: A Survey
RIFT: Multi-Modal Image Matching Based on Radiation-Variation Insensitive Feature Transform
Remote Sensing Image Matching Based on Adaptive Binning SIFT Descriptor
MODS: Fast and robust method for two-view matching☆
References
Distinctive Image Features from Scale-Invariant Keypoints
Object recognition from local scale-invariant features
Distinctive Image Features from Scale-Invariant Keypoints
A Combined Corner and Edge Detector
SURF: speeded up robust features
Related Papers (5)
Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography
Frequently Asked Questions (15)
Q2. How does the degraded orientation histogram achieve invariance?
The degraded orientation histogram constrains the gradient orientation from 0 to π, and then the histogram achieves invariance when the gradient orientation rotates by 180◦.
Q3. What is the main orientation assigned to each control point candidate?
A main orientation that is relative to the local gradient is assigned to each control point candidate before extracting the PIIFD.
Q4. How long does it take to register all 168 pairs of retinal images?
It takes approximately 41.3 min to register all 168 pairs of retinal images using their Harris-PIIFD algorithm (14.75 s per pair, standard deviation of 4.65 s).
Q5. How do the authors normalize the gradient magnitudes of each control point?
In a neighborhood surrounding each control point candidate, the authors normalize the first 20% strongest gradient magnitudes to 1, second 20% to 0.75, and by parity of reasoning the last 20% to 0.
Q6. What is the procedure for identifying PIIFDs?
PIIFDs are extracted relative to the main orientations of control point candidates therefore achieve invariance to image rotation, and a bilateral matching technique is applied to identify corresponding PIIFDs matches between image pairs (steps 2–4).
Q7. What is the importance of a reliable and fair evaluation method for measuring the performance of the retina?
A reliable and fair evaluation method is very important for measuring the performance since there is no public retinal registration dataset.
Q8. How many pairs of different-modal retinal images are selected?
In this test, 400 pairs of corresponding control points and 400 pairs of noncorresponding control points are chosen from 20 pairs of different-modal retinal images.
Q9. What is the way to register a poor quality multimodal image?
A robust local feature descriptor may bring to success the registration of poor quality multimodal retinal images, as long as it solves the following two problems: 1) the gradient orientations at corresponding locations in multimodal images may point to opposite directions and the gradient magnitudes usually change.
Q10. How can the authors extract the main orientation of each control point candidate?
Given the main orientation of each control point candidate (corner point extracted by Harris Detector), the authors can extract the local feature in a manner invariant to image rotation [42] and partially invariant to image intensity.
Q11. How many control point candidates are there in each experiment?
In their experiments, the average number of control point candidates is 231, the average number of initial matches (including incorrect matches) is 64.6, and the average number of final matches (after removing incorrect matches) is 43.2.
Q12. How many Harris corner points are detected in the PIIFD?
It has been confirmed that 200 Harris corner points are sufficient for subsequent processing, thus, in their experiments about 200 Harris corner points are detected by automatically tuning the sigma of Gaussian window.
Q13. How many bins are formed in the orientation histogram?
For a given small square in this neighborhood [e.g., the highlighted small square shown in Fig. 5(a)], an orientation histogram, which evenly covers 0◦– 360◦ with 16 bins (0◦, 22.5◦, 45◦,. . ., 337.5◦) is formed.
Q14. How long does it take to estimate the transformation parameters?
This task takes approximately 10 h, and afterward the authors develop a program to estimate the transformation parameters and overlapping percentage.
Q15. What is the scale factor of the proposed Harris-PIIFD algorithm?
8. The results of this experiment indicate that their proposed Harris-PIIFD can provide robust matching when the scale factor is below 1.8.