This paper demonstrates the application of Gabor Filter technique to enhance the fingerprint image. This work produces change in Gabor filter design by increasing the quality of an output which helps in higher security applications. The incoming signal in form of image pixel will be convoluted by Gabor filter to define the Edge and vale regions of fingerprint. The main characteristic of this paper is to store image pixel in memory if convolution signal is low and if the signal is high image is filtered.
Image processing is any form of signal processing for which the given input is an image file that is photo or video and the output generated is not necessary an image, it can also be a list of values or parameters depending on the requirement of the design. Most image-processing techniques involve treating the image as a two-dimensional signal.
ALGORITHM AND FLOWCHART
Initially, when the convolution signal is ‘0’ the input data which is in pixel format will enter the filter and stored in the memory. In ALU there is also a ROM which will store the Kernel coefficient. All the kernel coefficients are fixed values
When the convolution signal is triggered to ‘1’, then the convolution process starts and the convolution takes place between the memory and coefficients inside the MAC unit.
It takes 1 complete cycle to complete STEP 2, only one series of data will be convoluted at a time.
The design will count for 9 convolution operation before giving out the result of filtered image. Since there is 9 coefficient values it takes 9 consecutive cycles, which will result in output of the filter.
And if the convolution signal is ‘0’ the control will go to STEP 1 and if not the filtered output is obtained.
RESULTS AND DISCUSSION
A. Image output:
The input fingerprint image has been filtered and the clear output image has been obtained.
The waveform obtained from the convolution process done in MODELSIM.
C. Synthesis report:
This paper can be implemented in FPGA and the report of synthesis done in Spartan 6 family. Initially this design was implemented in Spartan 3E. It utilizes more memory and clock cycles and hence to avoid this issue we have implemented in Spartan 6 family. This produces considerable result and faster. This project can be further improvised by implementing on higher version boards.
D. Comparison of Device Used:
We have implemented this design in Xilinx 14.1 and the target device is xc6slx25t-3csg324 (Spartan 6). It was also checked in previous version Spartan 3E and the comparison is done.
This design can be effectively used for security and criminal investigation purposes. We can also implement this design in various fields like Biomedical, Defense. This design can be used in Acoustic application i.e to load audio databases.
Gabor Filter design for Fingerprint have been successfully completed. The area of the design has been significantly reduced but the function of the filter is perfectly maintained. The numbers of slices obtained from the previous design was reduced from 4488 slices to 384 with a utilization of 1%.
This design is used as one of the module in all security databases and since the world is getting worse in security, its very important to concentrate on personal security and since fingerprint differs for every individual, this gives a perfect individual security. In future this design can be taken into consideration and can be implemented in many real time applications as discussed above.
Source: Vellore Institute of Technology
Authors: R.Dhanabal | V.Bharathi | G.Prithvi Jain | Ganeash Hariharan | P.Deepan Ramkumar | Sarat Kumar Sahoo