Umarani Jayaraman

A Skin-Color and Template Based Technique for Automatic Ear Detection

This paper proposes an efficient skin-color and template based technique for automatic ear detection in a side face image. The technique first separates skin regions from non skin regions and then searches for the ear within skin regions. Ear detection process involves three major steps. First, Skin Segmentation to eliminate all non-skin pixels from the image, second Ear Localization to perform ear detection using template matching approach, and third Ear Verification to validate the ear detection using the Zernike moments based shape descriptor. To handle the detection of ears of various shapes and sizes, an ear template is created considering the ears of various shapes (triangular, round, oval and rectangular) and resized automatically to a size suitable for the detection. Proposed technique is tested on the IIT Kanpur ear database consisting of 150 side face images and gives 94% accuracy.

Indexing Multimodal Biometric Databases Using Kd-Tree with Feature Level Fusion

This paper proposes an efficient indexing technique that can be used in an identification system with large multimodal biometric databases. The proposed technique is based on Kd-tree with feature level fusion which uses the multi- dimensional feature vector. A multi dimensional feature vector of each trait is first normalized and then, it is projected to a lower dimensional feature space. The reduced dimensional feature vectors are fused at feature level and the fused feature vectors are used to index the database by forming Kd-tree. The proposed method reduces the data retrieval time along with possible error rates. The system is tested on multimodal databases (feature level fusion of ear, face, iris and signature) consists of 5400 images of 150 subjects (i.e. 9 images per subject per trait). Out of the 9, 8 images are used for training and 1 is used for testing. The performance of the proposed indexing technique has been compared with indexing based on score level fusion. It is found that proposed technique based on feature level fusion performs better than score level fusion.

An Indexing technique for biometric database

In this paper, an efficient indexing technique which can be used in an identification system with large biometric database has been proposed. The technique is based on the modified B+ tree which reduces the disk accesses and is found to be suitable for large biometric database. In this technique, first a multi-dimensional feature vector is projected to a lower dimensional feature space. Then, reduced dimensional feature vector is used to index the database by forming modified B+ tree. The proposed method decreases the data retrieval time along with possible error rates. This system is tested on Bath university and IITK iris databases with and without dimension reduction. It is observed that the system with reduced dimension performs almost equally well.

Ear Localization from Side Face Images using Distance Transform and Template Matching

The paper presents an efficient distance transform and template based technique for automatic ear localization from a side face image. The technique first segments skin and non-skin regions in the face and then uses template based approach to find the ear location within the skin regions. Ear detection proceeds as follows. First, edge map of the skin regions is computed and further processed to eliminate the spurious edges based on length and curvature based criterion. After getting the clean edge map, its distance transform is obtained on which ear localization process is carried out. Distance transform image of the edge map of an off-line created ear template is employed for ear localization. A Zernike moment based shape descriptor is used to verify the detections. The technique is tested on IIT Kanpur ear database which contains around 150 ear images and found to be giving 95.2% accuracy.

An efficient minutiae based geometric hashing for fingerprint database

Abstract This paper proposes an efficient indexing technique for fingerprint database using minutiae based geometric hashing. This technique consists of two stages, known as indexing and searching. For an accurate match at the time of searching, it has proposed a fixed length feature vector built from each minutia, known as Minutia Binary Pattern. Unlike any existing geometric based indexing technique, the proposed technique inserts each minutia along with the feature vector exactly once into a hash table. As a result, it reduces both computational and memory costs. Since minutiae of all fingerprint images in the database are found to be well distributed into the hash table, no rehashing is required. Experiments over FVC 2004 datasets prove the superiority of the proposed indexing technique against well known geometric based indexing techniques.

Connected component based technique for automatic ear detection

This paper presents an efficient technique for automatic ear detection from side face images. The proposed technique detects ear by exploiting its inherent structural details and is rotation, scale and shape invariant. It can detect ear without any training or assuming prior knowledge of the input image. The technique is based on connected component analysis of a graph constructed using the edge map of the image and is evaluated on a data set consisting of 2361 side face images collected at IIT Kanpur. Ear detection results are found to be very good and speak for the efficiency and robustness of the technique. To show the accuracy of the detection, detected ears are used for recognition and results are compared with the same obtained when ear cropping in done manually.

A score based indexing scheme for palmprint databases

This paper proposes an efficient retrieval technique which uses a new indexing scheme for a large palmprint database. For a given palmprint query image, the proposed indexing scheme makes use of match score vectors to determine an index and uses this index to reduce the search space of the database. Finally, the retrieval technique uses palmprint texture to find the top t best matches from the reduced search space. The proposed technique has been tested on publicly available palmprint database viz PolyU [1] of 7752 images and is found to be an efficient technique based on score vector indexing scheme. The test reveals that hitrate of 98.01% is acheived for the bin-success rate of 98.28%. The proposed indexing scheme is found to be more powerful than any other known indexing scheme based on match score vectors.

An efficient color and texture based iris image retrieval technique

This paper proposes a hierarchical approach to retrieve an iris image efficiently from for a large iris database. This approach is a combination of both iris color and texture. Iris color is used for indexing and texture is used for retrieval of iris images from the indexed iris database. An index which is determined from the iris color is used to filter out the images that are not similar to the query image in color. Further, iris texture features of those filtered images, are used to determine the images which are similar to the query image. The iris color information helps to design an efficient indexing scheme based on color indices. The color indices are computed by averaging the intensity values of all red and blue color pixels. Kd-tree is used for real-time indexing based on color indices. The iris texture features are obtained through Speeded Up Robust Features (SURF) algorithm. These features are used to get the querys corresponding image at the top best match. The performance of the proposed indexing scheme is compared with two well known iris indexing schemes (Mehrotra, Majhi, & Gupta, 2010; Puhan & Sudha, 2008) on UPOL (Dobes, Machala, Tichavsky, & Pospisil, 2004) and UBIRIS (Proencca & Alexandre, 2005) iris databases. It is observed that combination of iris color and texture improves the performance of iris recognition system.

An efficient technique for indexing multimodal biometric databases

This paper proposes an efficient indexing technique which can be used in an identification system with large multimodal biometric database. In this technique, multi-dimensional feature vectors of each trait (iris, signature, ear and face) are normalised and projected to a lower dimensional feature space. The reduced feature vectors are fused at feature level and used to index the database by forming Kd-tree. The performance of the proposed technique is also analysed with the feature vectors of all traits by first fusing them and projecting the fused feature vector to a lower dimensional space, and using it for indexing. Performance is also compared with the indexing based on score-level fusion. The experiment is performed on a multimodal database consisting of 5400 images of 150 subjects (i.e. nine images per subject, per trait). Out of the nine, eight images are used for training and one is used for testing. Our experiment shows that the proposed technique significantly reduces the data retrieval time along with possible error rates.

An iris retrieval technique based on color and texture

This paper proposes an efficient indexing scheme that can be used for retrieval from a large iris database. For a given color iris query image, the proposed indexing scheme makes use of iris color to determine an index and uses this index to reduce the search space in the large iris database. Further, for query q, the retrieval technique uses iris texture to find the top best match from the reduced search space. The proposed technique has been tested on two publicly available color iris databases, viz UPOL [10] of 384 images and UBIRIS [13] of 1860 fully noisy images and is found to be robust against change in gaze, illumination, partial occlusions and scale. In both the databases, the test reveals that a small subspace is sufficient to achieve 100% hitrate for the top best match under various scales, illumination and partial occlusion. The performance of the proposed indexing scheme is analyzed against the group based color indexing scheme proposed in [14]. The results show that proposed indexing scheme is performing better as compared to group based color indexing scheme with respect to hitrate, penetration rate and CMC curve.