Orientation Field Estimation for Latent Fingerprint Enhancement
read more
Citations
Segmentation and Enhancement of Latent Fingerprints: A Coarse to Fine Ridge Structure Dictionary
Deep convolutional neural network for latent fingerprint enhancement
Latent orientation field estimation via convolutional neural network
Learning Fingerprint Reconstruction: From Minutiae to Image
Localized Dictionaries Based Orientation Field Estimation for Latent Fingerprints
References
Handbook of Fingerprint Recognition
Fingerprint image enhancement: algorithm and performance evaluation
Digital Image Processing 3rd Edition
Markov Random Field Modeling in Image Analysis
Techniques for automatically correcting words in text
Related Papers (5)
Fingerprint image enhancement: algorithm and performance evaluation
Segmentation and Enhancement of Latent Fingerprints: A Coarse to Fine Ridge Structure Dictionary
Frequently Asked Questions (15)
Q2. What is the proposed orientation field estimation algorithm?
The proposed orientation field estimation algorithm consists of an off-line dictionary construction stage and an on-line orientation field estimation stage.
Q3. What is the compatibility constraint between two neighboring patches?
Due to the fact that relatively large size orientation patches are treated as a whole and adjacent patches contain an overlapping region, the compatibility constraint holds in both low curvature regions as well as high curvature regions (such as core and delta).
Q4. How many rolled fingerprints were used as the background database?
To make the latent matching problem more realistic and challenging, 27,000 rolled fingerprints (file fingerprints) in the NIST SD14 database were used as the background database.
Q5. How many reference orientation patches are used in the dictionary?
When the size of the patch is 10×10 blocks and 50 reference orientation fields are used, the number of reference orientation patches is around 23K.
Q6. What are the main reasons why automatic fingerprint extraction is desirable?
Automatic latent feature extraction is desirable for several reasons.1) Reducing the time spent by latent examiners in manual markup.
Q7. Why do latent fingerprints need to be manually marked?
Because of the poor image quality, features (such as minutiae) in latents need to be manually marked by latent examiners so that they can be searched against large fingerprint databases by automated fingerprint identification systems (AFIS).
Q8. How did the authors obtain the overlapped latent fingerprints?
These overlapped latent fingerprints were obtained using the following procedure: 1) press two fingers at roughly the same location on a white paper, 2) enhance the latent prints using black powder and brush, and 3) convert the enhanced prints into electronic version using a general purpose scanner.
Q9. How many orientation patches are available in the dictionary?
A number of orientation patches, whose orientation elements are all available, are obtained by sliding a window (whose size is b×b blocks) across each reference orientation field and its mirrored version.
Q10. What is the proposed algorithm for estimating the orientation field of a fingerprint?
1) The initial orientation field estimation algorithm detects one dominant orientation element in the nonoverlapped fingerprint region and two dominant orientation elements in the overlapped region.
Q11. What is the performance of the proposed algorithm?
Given prior knowledge of fingerprint structure, which is represented by a dictionary of reference orientation patches and compatibility constraints between adjacent orientation patches, the proposed algorithm obtains better performance for latents than published algorithms (see Fig. 2).
Q12. how many neighbors are viewed as candidates for replacing the noisy initial orientation patch?
For each initial orientation patch, its six nearest neighbors in the dictionary are viewed as candidates for replacing the noisy initial orientation patch.
Q13. how many orientation fields are generated from a set of high quality fingerprints?
The orientation fields (defined on blocks of size 16×16 pixels) of these fingerprints are estimated using56 a state-of-the-art algorithm, VeriFinger 6.2 SDK [40].
Q14. What are the two types of algorithms used to regularize the noisy orientation field?
To deal with this problem, two types of algorithms have been adopted to regularize the noisy orientation field, namely, orientation field smoothing and global parametric model fitting.
Q15. How many reference orientation patches in the dictionary?
The number of reference orientation patches in the dictionary depends on the number of reference orientation fields and the size of the patch.