Artificial Neural Network Based Iris Recognition
Individual recognition using Iris is most commonly employed in all place. This requires some special cameras to take the Iris image and the obtained iris images were classified based on the features extracted. Only identification of person is not the application of Iris recognition. It can be de
2025-06-28 16:25:10 - Adil Khan
Artificial Neural Network Based Iris Recognition
Project Area of Specialization NeuroTechProject SummaryIndividual recognition using Iris is most commonly employed in all place. This requires some special cameras to take the Iris image and the obtained iris images were classified based on the features extracted. Only identification of person is not the application of Iris recognition. It can be developed to do many process. The iris image captured from the cameras were taken. Individual recognition using Iris is most commonly employed in all place. In the proposed approach the process of recognition of the persons based on Iris image is employed. The features used in this approach are based on the contour of the iris-pupil boundary obtained from Euclidean distance formula functions and is named eye signature.
Multilayer Feedforward Neural Network (MFNN) is used as classification model to perform iris recognition using eye signature feature vector. The feedforward neural networks have not been applied before on eye signatures. Since eye signature is 1D, it simplifies the structure of neural networks. Hence this method has lesser complexity than the existing neural networks based techniques. The performance of the process is measured based on the performance metrics. The performance of the process measured indicates that the used approach is more improved compared to the other existing approaches for the iris recognition process.
Project ObjectivesThe objective of this project is to do Iris recognition and classification using inner iris boundary which is iris-pupil boundary. Images which are used for training and testing of ANN are from CASIA Iris database Version 1 . To achieve this objective first of all ROI must be detected which is pupil region. Then contour of that region must be formed. This contour should be used to find distances from center to all points on the contour which will make the feature vector. Finally use this feature vector to train ANN. The main aim is to achieve zero failure rate with 30, 40, and 50 dB SNR injected training data.
Project Implementation MethodThree Stage Approach:
1. Iris image preprocessing
- Filtering the image to remove unwanted portions and noise from it
2. Feature Extraction
- Pupil localization.
- Localizing the inner boundary of the iris.
- Locating the contour of interior (Pupil-Iris) boundary of the iris.
- Concatenating the four quadrant vectors found in previous step to obtain feature vector, named as Eye Signature.
- Implement min-max normalization or standardization for feature scaling.
3. Classification using NN
- Employ a Robust backpropagation algorithm with MFNN
A brilliant technique is used for representing the iris using contour of inner (Pupil-Iris) boundary of the iris. The classification is performed using MLP MFNN via backpropagation learning. The method used is lesser complex and more efficient than the existing neural networks based techniques
Technical Details of Final DeliverableThe first step is to locate the inner boundary of iris which is pupil-iris boundary. This can be done by locating the pupil region. Then employing filters to remove unwanted area. Morphological filters are suitable for this purpose. Median filter is useful for removing noise. This type of filter doesn’t affect strong edges. Edge detection technique is used to detect the contour of inner boundary of iris. “Canny method” is used to detect edges in this project. This method finds gradient of local maxima to find the edges. The method detects both strong and weak edges by using two thresholds but weak edge is only included if it is connected to a strong edge.
Next step is to find the inner boundary of iris. To accomplish this the center of pupil is chosen as the reference point. The pupil-iris boundary is represented as binary ones. In this way all the points on boundary can detected by searching for one in all direction. Then calculate the distance between the points on boundary and the center of iris to form feature vector. The name of feature vector is “Eye Signature”.
The eye signatures extracted using the method discussed above are used for classification. For this purpose a Multi-Feed Forward Neural Network is used.
Final Deliverable of the Project HW/SW integrated systemCore Industry HealthOther IndustriesCore Technology NeuroTechOther TechnologiesSustainable Development Goals Quality EducationRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 32000 | |||
| matlab software license | Equipment | 1 | 12000 | 12000 |
| screen | Equipment | 1 | 10000 | 10000 |
| Project Report Printing+ poster priting | Miscellaneous | 1 | 10000 | 10000 |