Post on 24-Feb-2016
description
By Pushpita Biswas
Palm print Verification for Controlling Access to Shared Computing Resources
Under the guidance of Prof. S.Mukhopadhyay and Prof. P.K.Biswas
Why access security is used?
Why Palm print verification?
1. no need to memorize codes or passwords.
2. more reliable
Four Stages of Palm print Verification
Image acquisitionPalm positioning Feature extraction Palm print matching
Image acquisition
Palm Positioning
Feature extraction
Register or
verify?
Palm print matching
TIFF file (gray scale)
Gray-scale Image
Line edge map
Verify
Decision
Register
Registered model
Database
Flow Chart
1. Image acquisition
Image of the user’s hand is taken via a camera and stored a grayscale TIFF file.
2. Palm positioning
Boundary extraction and edge thinning Feature point location Establishment of coordinate system Sub image normalization
Boundary extraction and edge thinning
1. Gradient magnitude of each pixel computed using set of sobel masks for detecting horizontal, vertical and diagonal edges.
2. Adaptive thresholding :- Gr => highest gradient value taken as referenceRatio_Gradient => predetermined constant between 0 and 1 T_Gradient => Threshold value
T_Gradient = Gr * Ratio_Gradient3. Selected pixels removed from binary image to reduce all lines
in the image to a single pixel width.
Feature point locationIn the boundary image’s line pattern the bottom of a valley is a short curve joining the edges of adjacent fingers.The key points are best represented as those curve’s midpoints.Establishment of
the coordinate system
The x-axis passes through K1 and K3.The y-axis is perpendicular to the x-axis and passes through K2
1. Sort the parallel line pairs, so that the line pairs are stored in left to right order. 2. For each parallel pair Pi in the sorted array, form a V- shape pair with the right edge of Pi and the left edge of Pi+1 (i = 0..I-2, where I is the total number of parallel pairs)
Sub image normalization
The rectangle specifications :1.distance between x-axis and
rectangle’s nearest side isRefLength * 0.25,
RefLength =>distance between K1 and K3
2.sides parallel to x-axis and y-axis3.symmetric with respect to y-axis4.sides have length of RefLength
Scaling and rotation is followed
3. Feature extraction
Image PreprocessingA 3*3 averaging mask is used, which smoothes the image and minimizes the noise impact.
Line DetectionStandard Sobel edge detector is used followed by thresholding on edge magnitude.
Image ThresholdingThreshold value calculated on basis of a percentage of image area.
Line thinningResulting image contains lines of only a single pixel width
Results
Results
Next
Thresholding of two sample images, of same person captured under different
lighting conditions
Return
Result of line detection
Return
Thinning and straight line approximation
Result of thinning Result of Line approximation
Contour tracing and the Dynamic Two-Strip (DYN2S) algorithm is applied to establish a set of straight line
segments that approximate the extracted palm print lines.
4. Palm print matching1. Line segment Hausdorff distance (LHD) is
applied. m and t are 2 line segments
Angle distance by tangent function with respect to smallest angle between m and t.Predetermined weight of angle distance
2. Decision Making
Choice of method depends on system specification
Results for palm print matching system.Thus Threshold value is decided.
Conclusion
The system will work well on images with a uniform background, but this can be further extended to handle images with arbitrary backgrounds. Since the algorithm for locating and aligning the palm print is based on line detection instead of simple segmentation, makes the system more robust and suitable for security applications with outdoor cameras.
References
M.K.Leung, A.C.M. Fong, Siu Cheung Hui “Palm print Verification for Controlling Access to Shared Computing Resources,” IEEE Pervasive Computing, vol. 6, no. 4, 2007, pp. 40–47.
W.J. Rucklidge, “Efficiently Locating Objects Using the
Hausdorff Distance,” Int’l J. Computer Vision, vol. 24, no. 3, 1997,pp. 251–270.
M.K. Leung and Y.H. Yang, “Dynamic Two-Strip Algorithm in Curve Fitting,” Pattern Recognition, vol. 23, nos. 1–2, 1990, pp. 69–79.
Thank you