Tanuja K. Sarode

An Efficient Fast Algorithm to Generate Codebook for Vector Quantization

In this paper we present a very simple and yet effective algorithm to generate codebook. The algorithm uses sorting method to generate codebook and the codevectors are obtained by using median approach. The proposed algorithm was experimented on six different images each of size 512 x 512 and four different codebooks of sizes 128, 256, 512 and 1024 are generated. The proposed algorithm is found to be much faster than the LBG and KPE algorithm. The performance of this algorithm is better than LBG and KPE algorithms considering MSE, PSNR and execution time. The proposed algorithm gives less MSE as compared to the LBG for the codebooks of sizes 128, 256, 512 & 1024 respectively. It also gives higher PSNR as compared to LBG for the codebooks of various sizes.

Improved texture feature based image retrieval using Kekre’s fast codebook generation algorithm

Technological advances in digital imaging, broadband networking, and data storage have motivated people to communicate and express by sharing images, video, and other forms of media online [1]. Although the problems of acquiring, storing and transmitting the images are well addressed, capabilities to manipulate, index, sort, filter, summarize, or search through image database lack maturity. Modern image search engines [1–3, 30] retrieve the images based on their visual contents, commonly referred to as Content Based Image Retrieval (CBIR) systems. CBIR systems have found applications in various fields like fabric and fashion design, interior design as panoramic views [5, 20, 38–41], art galleries [22], museums, architecture/ engineering design [22], weather forecast, geographical information systems, remote sensing and management of earth resources [40, 41], scientific database management, medical imaging, trademark and copyright database management, the military, law enforcement and criminal investigations [41], intellectual property, picture archiving and communication systems, retailing and image search on the Internet (very general image content).

Image retrieval by Kekre's transform applied on each row of Walsh transformed VQ codebook

The new technique for image retrieval using the color-texture features extracted from images based on vector quantization with Kekres fast codebook generation is proposed. This gives better discrimination capability for CBIR. Here the database image is divided into 2x2 pixel windows to obtain 12 color descriptors per window (Red, Green and Blue per pixel) to form a vector. Collection of all such vectors is a training set. Then the Kekres Fast Codebook Generation (KFCG) is applied on this set to get 16 codevectors. The Walsh transform is applied on each column of the codebook, followed by Kekres transform applied on each row of the Walsh transformed codebook. This transform vector then is used as the image signature (feature vector) for image retrieval. The method takes lesser computations as compared to conventional Walsh applied on complete image. The method gives the color-texture features of the image database at reduced feature set size. Proposed method gives better precision and recall as compared to full Walsh based CBIR. Proposed method avoids resizing of images which is required for any transform based feature extraction method.

Bi-Level Vector Quantization Method for Codebook Generation

Vector Quantization is lossy data compression technique and has various applications. Key to Vector Quantization is good codebook. Once the codebook size is fixed then for any codebook generation algorithm the MSE reaches a value beyond which it cannot be reduced unless the codebook size is increase. In this paper we are proposing bi-level codebook generation algorithm which reduces mean squared error (MSE) for the same codebook size. For demonstration we have used codebooks obtained from well known Linde Buzo and Gray (LBG) algorithm. The proposed method is general and can be applied to any codebook generation algorithm.

Image Segmentation of Mammographic Images Using Kekre'S Proportionate Error Technique on Probability Images

Mammography is well known method for detection of breast tumors. Early detection and removal of the primary tumor is an essential and effective method to enhance survival rate and reduce mortality. In this paper, proposed algorithm uses probability of mammographic image as input for vector quantization .For region forming Kekres Proportionate Error (KPE) algorithm is used and codebook of size 128 is formed .Further this 128 clusters were used for region merging using KPE algorithm for reclustering .To separate tumor ,post processing is done by morphological operations. For this tumor sectional area is calculated and center point is compared with LBG algorithm for segmentation of mammographic images.

Iris Recognition Using Vector Quantization

In today’s world, where terrorist attacks are on the rise, employment of infallible security systems is a must. Iris recognition enjoys universality, high degree of uniqueness and moderate user co-operation. This makes Iris recognition systems unavoidable in emerging security & authentication mechanisms. We propose an iris recognition system based on vector quantization. The proposed system does not need any pre-processing and segmentation of the iris. We have tested LBG, Kekre’s Proportionate Error Algorithm (KPE) & Kekre’s Fast Codebook Generation Algorithm (KFCG) for the clustering purpose. From the results it is observed that KFCG requires 99.79% less computations as that of LBG and KPE. Further the KFCG method gives best performance with the accuracy of 89.10% outperforming LBG that gives accuracy around 81.25%. Performance of individual methods is evaluated and presented in this paper.

Identification of multi-spectral palmprints using energy compaction by Hybrid wavelet

Orthogonal Transforms and Wavelets can be used to extract features of a biometric in frequency domain. They also exhibit the property of energy compaction which can be used to select few coefficients as features of an image. Here we make use of a Hybrid wavelet, generated by using Kronecker product of two existing orthogonal transforms, Walsh and DCT to identify multi-spectral palmprints. A threshold energy value is chosen to select all coefficients whose total energy is above that value. One-to-many identification on a large database containing 3 sets of 6000 multi-spectral palmprint images from 500 different palms is used to validate the performance of the proposed method and matching accuracy in terms of genuine acceptance ratio of 99.979% using score level fusion has been obtained. Hence this method can significantly improve the identification rates for palmprint images.

New Clustering Algorithm for Vector Quantization using Walsh Sequence

In this paper we present an effective clustering algorithm to generate codebook for vector quantization (VQ). Constant error is added every time to split the clusters in LBG, resulting in formation of cluster in one direction which is 135 in 2dimensional case. Because of this reason clustering is inefficient resulting in high MSE in LBG. To overcome this drawback of LBG proportionate error is added to change the cluster orientation in KPE. Though the cluster orientation in KPE is changed, its variation is limited to ± 45 over 135. KEVR introduces new orientation every time to split the clusters. But in KEVR the error vector sequence is the binary representation of numbers, so the cluster orientation change slowly in every iteration. To overcome this drawback we propose the technique which uses Walsh sequence to rotate the error vector. The proposed technique (Kekre’s error vector rotation using Walsh – KEVRW) is based on KEVR algorithm. The proposed methodology is tested on different training images for code books of sizes 128, 256, 512, 1024. Our result shows that KEVRW gives less MSE and high PSNR compared to LBG, KPE and KEVR. General Terms Image Processing, Vector Quantization, Data Compression.

Fingerprint Identification using Sectorized Cepstrum Complex Plane

Using biometrics to verify a person’s identity has several advantages over the present practices of personal identification numbers (PINs) and passwords. Minutiae-based automated fingerprint identification systems are more popular, but they are more computationally complex and time consuming. In this paper we propose a simple yet effective technique for fingerprint identification. This method is image-based in which feature vectors of a fingerprint are extracted after sectorization of the cepstrum of a fingerprint. They are matched with those stored in the database. The experimental results show that this algorithm could correctly identify fingerprints with accuracy more than 96% in case of larger number of sectors.

Transform based face recognition with partial and full feature vector using DCT and Walsh transform

This paper presents, efficient transform based face recognition technique which considers full and partial feature vector of an image. 2D-DCT and Walsh transform is applied on the resized image of size 128x128, to obtain its feature vector. Partial feature vector is obtained by selecting 75% rows and columns of feature vector, 50% rows and columns of feature vector and so on. The smallest size of partial feature vector is selected as 4x4. Proposed technique is tested on two different databases. Georgia Tech Face Database contains JPEG color images and Indian Face Database contains Bitmap color images of varying size. Recognition rate is calculated for varying size of selected feature vector using DCT and Walsh transform and compared. Also computational complexity in terms of number of CPU units is compared in both the cases: with full feature vector and with partial feature vector. Results show that, Walsh transform gives better recognition rate than DCT and number of CPU units required using 2D- Walsh transform is almost 9 times less than that of required by using 2D-DCT. This is because the multiplications required in Walsh transform are zero.