Dr D Madhavi

Scopus Publications

An Optimized Deep Features for Detecting Tampered Region from the Copy Move Forgery Image

Unsupervised Content Mining in CBIR: Harnessing Latent Diffusion for Complex Text-Based Query Interpretation
Venkata Rama Muni Kumar Gopu and Madhavi Dunna
MDPI AG
The paper demonstrates a novel methodology for Content-Based Image Retrieval (CBIR), which shifts the focus from conventional domain-specific image queries to more complex text-based query processing. Latent diffusion models are employed to interpret complex textual prompts and address the requirements of effectively interpreting the complex textual query. Latent Diffusion models successfully transform complex textual queries into visually engaging representations, establishing a seamless connection between textual descriptions and visual content. Custom triplet network design is at the heart of our retrieval method. When trained well, a triplet network will represent the generated query image and the different images in the database. The cosine similarity metric is used to assess the similarity between the feature representations in order to find and retrieve the relevant images. Our experiments results show that latent diffusion models can successfully bridge the gap between complex textual prompts for image retrieval without relying on labels or metadata that are attached to database images. This advancement sets the stage for future explorations in image retrieval, leveraging the generative AI capabilities to cater to the ever-evolving demands of big data and complex query interpretations.

RCVM-ASS-CICSKA-PAPT-VDF: VLSI design of high-speed reconfigurable compressed Vedic PAPT-VDF filter for ECG medical application
K. V. Suresh Kumar and D. Madhavi
Wiley
AbstractDuring signal acquisition, the signals are impacted by multiple noise sources that must be filtered before any analysis. However, many different filter implementations in VLSI are dispersed among many studies. This study aims to give readers a systematic approach to designing a Pipelined All‐Pass Transformation based Variable digital filter (PAPT‐VDF) to eliminate the high‐frequency noise from ECG data. The modified design emphasizes first‐ and second‐order responses to obtain high‐speed filter realization with high operating frequencies. The addition of adder and multiplier designs to the hardware architecture of a filter design improves performance. The fundamental blocks of the filter design are the adder and multiplier. The adder and multiplier are employed with an Adaptable stage size‐based concatenation, incremented carry‐skip adder (ASS‐CICSKA), and Improved reconfigurable compressed Vedic multiplier (IRCVM). Utilizing the adder design diminishes the delay with enhanced performance because receiving the carry from an incrementation block is not mandatory. In the multiplier design, the compressor and the reconfigurable approach are adapted with a data detector block to detect the redundant input and lower the logic gates' switching activity with less area overhead. The proposed filter design is implemented in vertex 7 FPGA family device, and the performance measures are analyzed regarding area utilization, delay, power, and frequency. Also, by using the denoised signal, the mean square error (MSE), and signal‐to‐noise ratio (SNR) are evaluated in the MATLAB platform.

Zero-Shot Sketch-Based Image Retrieval Using StyleGen and Stacked Siamese Neural Networks
Venkata Rama Muni Kumar Gopu and Madhavi Dunna
MDPI AG
Sketch-based image retrieval (SBIR) refers to a sub-class of content-based image retrieval problems where the input queries are ambiguous sketches and the retrieval repository is a database of natural images. In the zero-shot setup of SBIR, the query sketches are drawn from classes that do not match any of those that were used in model building. The SBIR task is extremely challenging as it is a cross-domain retrieval problem, unlike content-based image retrieval problems because sketches and images have a huge domain gap. In this work, we propose an elegant retrieval methodology, StyleGen, for generating fake candidate images that match the domain of the repository images, thus reducing the domain gap for retrieval tasks. The retrieval methodology makes use of a two-stage neural network architecture known as the stacked Siamese network, which is known to provide outstanding retrieval performance without losing the generalizability of the approach. Experimental studies on the image sketch datasets TU-Berlin Extended and Sketchy Extended, evaluated using the mean average precision (mAP) metric, demonstrate a marked performance improvement compared to the current state-of-the-art approaches in the domain.

Detection and Classification of Diabetic Retinopathy Through Identification of Blood Vessel Thickness Using FOFF& ML Classifiers

Identification of Microaneurysms and Exudates for Early Detection of Diabetic Retinopathy
G Indira Devi and D. Madhavi
The Science and Information Organization
—Diabetic retinopathy (DR) is a condition that may be a complication of diabetes, and it can damage both the retina and other small blood vessels throughout the body. Microaneurysms (MA’s) and Hard exudates (HE’s) are two symptoms that occur in the early stage of DR. Accurate and reliable detection of MA’s and HE’s in color fundus images has great importance for DR screening. Here, a machine learning algorithm has been presented in this paper that detects MA’s and HE’s in fundus images of the retina. In this research a dynamic thresholding and fuzzy c mean clustering with characteristic feature extraction and different classification techniques are used for detection of MA’s and HE’s. The performance of system is evaluated by computing the parameters like sensitivity, specificity, accuracy, and precision. The results are compared between different types of classifiers. The Logistic Regression classifier (LRC) performance is good when compared with other classifiers with an accuracy of 94.6% in detection of MA’s and 96.2% in detection of HE’s.

Stacked Siamese Neural Network (SSiNN) on Neural Codes for Content-based Image Retrieval
Gopu V. R. Muni Kumar and D. Madhavi
Institute of Electrical and Electronics Engineers (IEEE)
Content-based image retrieval (CBIR) represents a class of problems that aims at finding relevant images in response to an image-based search query. The CBIR systems use similarity measures or distance metrics between a group of representative features in the query image and those in the image repository. Traditionally, these features were generated by hand, employing image features such as colour, texture, shape, and so on. Due to the fact that these methods do not provide a comprehensive perspective of the images, they cannot be widely utilized in contemporary CBIR systems. This is due to the so-called semantic gap between query intent and system perspective. The most recent advancements in deep learning offer a viable alternative to manually built features, leveraging the representational learning capability of deep neural networks. This paper presents a method of implementing a CBIR system using a multi-stage approach known as classify, differentiate, and retrieve (CDR). The first stage involves using a deep neural network to encode the images. Later, a custom-trained stacked Siamese Neural network (SSiNN) is employed to differentiate the latent space representation of the images obtained from the first stage. The experimental results for the CIFAR-10 dataset were presented, along with an algorithm for applying this strategy to any generic dataset. Experimental outcomes demonstrate that the proposed strategy is superior to the current best practices.

OPTIMIZATION of QBIC SYSTEMS USING LG WAVELETS

Optimization of log gabor filters using genetic algorithm for query by image content systems
N. Jyothi, D. Madhavi, and M. R. Patnaik
Springer Singapore
Query By Image Content (QBIC) system permits the user to extract similar images from the large database by specifying the query image. In this paper, a new hybrid QBIC system is proposed, which effectively utilizes the color, texture, and shape features by sequential process of three pipe stages which gives good retrieval performance and also reduces computational complexity. In first pipe stage, color features are extracted with histograms based on HSV color space. In second pipe stage, texture information is obtained by optimizing log Gabor filters using genetic algorithm. Finally, in third pipe stage, shape features are extracted with signature functions using polygonal algorithm. Simulation results show that the proposed hybrid sequential pipe stage combination gives good performance metrics in terms of mean recall and mean precision and also reduces the number of filters compared to the existing algorithms.

PSO optimized log gabor QBIC system
Blue Eyes Intelligence Engineering and Sciences Engineering and Sciences Publication - BEIESP
In modern years, there is substantially technical progression in research area pertaining to image retrieval, in specific Query By Image Content (QBIC) system. It has turned out to be essential to deliver adept and effective method to retrieve images from the gigantic collections of images utilized in heterogeneous applications. In this paper, a hybrid QBIC retrieval system known to be PSO optimized Log Gabor QBIC system that retrieves color features, texture features and shape features of the images in three consecutive stages has been developed. In the proposed system, color features are retrieved by means of color histogram in the first stage. In subsequent stage, the texture features are extracted by tuning Log Gabor filters using Particle Swarm Optimization(PSO). Lastly, shape features are retrieved by polygonal fitting algorithm. The recommended method displays enhanced retrieval rate in terms of mean recall and mean precision when compared to the prevailing standard systems.

A hybrid content based image retrieval system using log-gabor filter banks
D. Madhavi, Khwaja Muinuddin Chisti Mohammed, N. Jyothi, and M. Ramesh Patnaik
Institute of Advanced Engineering and Science
<p>In this paper, a new efficient image retrieval system using sequential process of three stages with filtering technique for the feature selection is proposed. In the first stage the color features are extracted using color histogram method and in the second stage the texture features are obtained using log-Gabor filters and in the third stage shape features are extracted using shape descriptors using polygonal fitting algorithm. The proposed log-Gabor filter in the second stage has advantages of retrieving images over regular Gabor filter for texture. It provides better representation of the images. Experimental evaluation of the proposed system shows improved performance in retrieval as compared to other existing systems in terms of average precision and average recall.</p>

Genetic Algorithm-Based Optimized Gabor Filters for Content-Based Image Retrieval
D. Madhavi and M. Ramesh Patnaik
Springer Singapore
Fast and exact searching of digital image from the large database is the great demand. In this paper, a hybrid technique to improve the efficiency of content-based image retrieval (CBIR) is proposed. It uses combination of color, texture, and shape feature extraction methods. Color features are extracted using HSV histograms. For texture feature extraction, instead of traditional Gabor filter, four Gabor filters are simultaneously tuned in the desired direction using genetic algorithm and features are extracted in each direction simultaneously. The shape features are obtained using shape signature function with polygonal fitting algorithm. By the sequential process of these three stages, the retrieval performance is greatly improved. The simulation results prove that the proposed analysis gives significant improvement with respect to retrieval performance and computational complexity with the other proposed schemes.

Implementation of Non Linear Companding Technique for Reducing PAPR of OFDM
D. Madhavi and M. Ramesh Patnaik
Elsevier BV
Abstract High Peak-to-Average Power Ratio (PAPR) of transmitted signal is one of the foremost problems to implement Orthogonal Frequency Division Multiplexing (OFDM) System. To transform the OFDM signal into anticipated statistics form that are identified by a linear piecewise function, the new Non-Linear Companding transform (NCT) set of rules is used. The variable slopes and inflexion points are introduced inside the probability density function (PDF) while the PAPR and Bit Error Rate (BER) are compared to achieve efficiency in the overall performance and flexibility in the Non-Linear Companding (NCL) form. The expected transform gain and signal attenuation factor and all the theoretical value study of this set of rules are given. The main parameters are evaluated specifically based on the selection criteria of the transform parameters focusing on robustness and execution aspects. The exploration is exactly proved in Simulink.

Image retrieval based on tuned color gabor filter using genetic algorithm

Implementation of heartbeat sensing using PSoC3
Ramesh Babu Chukka, D. Madhavi, N. Jyothi, and Ch Sumanth Kumar
Springer Singapore
Electrocardiogram is a register of heart’s electrical activity. A wide range of heart conditions can be interpreted using ECG. It is increasingly used in medical sciences and technologies as a valuable medical diagnostic tool. For interpretation of ECG, amplifiers play an important role for such instrumentation systems. There is a growing demand for affordable, portable ECG machine. So by choosing the appropriate components suitable for portable applications, portable ECG machines can be developed. The objective is to develop a 3-lead portable feasible user-friendly and economical ECG system which can be managed by a common man.

Development of 3D model with ISO surface reconstruction algorithm in cosmetic surgical aplications

Image retrieval using GA optimized gabor filter
D. Madhavi and M. Ramesh Patnaik
Indian Society for Education and Environment
Objective: A Hybrid content based image retrieval method is proposed in this paper. This method extracts color, tuned texture and shape features of the images in three successive phases. Methodology: In proposed system, color features are extracted using color histogram method in the first phase. The tuned texture features are extracted by employing GA optimized Gabor filters in second phase. Finally, shape features are extracted using the polygonal fitting algorithm. The best match output images of each phase are given as input images to the next phase to obtain ‘S’ best match images out of ‘N’ database images. Findings: The novelty of proposed system is that it employs a tunable filter that is tuned with the query image dynamically. The tuning of Gabor filter is implemented using GA in second phase. The proposed method shows improved retrieval rate in terms of average recall and average precision compared to the existing systems. The computation complexity is also found to be less than other existing methods. Applications: It can be employed in numerous fields such as medical, satellite, multimedia, and surveillance imaging systems, etc. where the retrieval of related images from huge databases is critical task for analysis.

Block based partial update NLMS algorithm for adaptive decision feedback equalization
Ch. Sumanth Kumar, D. Madhavi, N. Jyothi, and K.V.V.S. Reddy
Elsevier BV
Abstract Decision feedback equalizers are commonly employed to reduce the intersymbol interference that is caused by the time dispersive channel.In this paper a block based partial update normalized LMS algorithm is proposed,which significantly reduces the computational complexity over the other LMS based algorithms.The important characteristic of this algorithm is that only a part of the filter coefficients are updated in every iteration.The frequency domain representation facilitates, easier to choose step size with which the proposed algorithm convergent in the mean squared sense, whereas in the time domain it requires the information on the largest eigen value of the correlation matrix of the input sequence. Simulation studies shows that the proposed realization gives good performance characteristic in terms of convergence rate.

Block and partial update sign normalized LMS based adaptive decision feedback equalization
Ch. Sumanth Kumar, D. Madhavi, N. Jyothi, and K. V. V. S. Reddy
IEEE
In this paper, a simple and efficient Block and partial update sign normalized LMS (BPSNLMS) algorithm is proposed for the decision feedback equalization. The proposed implementation is suitable for applications requiring long adaptive equalizers, as is the case in several high-speed wireless communication systems. The proposed algorithm yields good bit error rate performance over a reasonable signal to noise ratio. In this scheme the incoming data is partitioned into non overlapping blocks and the filtering operation has been performed in frequency domain with FFT (overlap and save method). In each iteration, only a part of the filter coefficients are updated so that it reduces the computational complexity and improves speed of operation. The signed versions of LMS algorithm makes equalization techniques multiplication free and facilitates efficient digital implementation using shift registers. The frequency domain representation facilitates, easier to choose step size with which the proposed algorithm convergent in the mean squared sense, whereas in the time domain it requires the information on the largest eigen value of the correlation matrix of the input sequence. Simulation studies shows that the proposed realization gives better performance compared to existing realizations in terms of convergence rate.

Transformation techniques for high speed implementation of recursive loop algorithms
D. Madhavi, N. Jyothi, Ch. Sumanth Kumar, P. L. H. Varaprasad, N. K. Surname, P. M. Surname, and A. S. A. Surname
IEEE
In digital communication systems, high speed transmission requires implementation of high speed digital circuits which include adaptive equalizers, encoders and other signal processing algorithms. This paper proposes several high level architectural transformations that can be used to design families of architectures for a given algorithm. It also deals with high-level algorithm transformations such as look ahead. It is applied to design pipelined adaptive digital filters and parallel recursive digital filters. The technique used here can be used for high speed, low area and low power implementations of DSP systems for various applications such as multimedia, wired and wireless communications. Examples of these algorithms are differential pulse code modulation (DPCM), adaptive differential pulse code modulation (ADPCM), decision feedback equalizers (DFE's). DSP algorithms are described using mathematical formulations at a higher level. The internal feedback in these structures makes it difficult to implement using either pipelining or parallel processing techniques. This paper proposes different computation approaches for quantization algorithms, which can be easily pipelined. These approaches are suitable for real time highspeed implementation of quantizer loop operations. The power consumption of the proposed systems is less by the use of tri-state buffers in the implementation.

A new sign normalized block based adaptive decision feedback equalizer for wireless communication systems
Ch. Sumanth Kumar, D. Madhavi, Rafi Ahamed Shaik, and K. V. V. S. Reddy
IEEE
Decision feedback equalizers are used in wireless and mobile communications to reduce the intersymbol interference that is caused by the time dispersive channel. Here, an adaptive decision feedback equalizer is presented with a new adaptive algorithm. The algorithm uses sign normalized block based least mean square algorithm, and achieves a significant reduction of computational complexity. The proposed algorithm yields good bit error rate performance over a reasonable signal to noise ratio. In this scheme the incoming data is partitioned into non overlapping blocks and the filtering operation has been performed in frequency domain with FFT(overlap and save method). In this the correction applied to the tap weight vector and it is normalized with respect to the squared euclidean norm of the tap input vector at time n. The frequency domain representation facilitates, easier to choose step size with which the proposed algorithm convergent in the mean squared sense, whereas in the time domain it requires the information on the largest eigen value of the correlation matrix of the input sequence. Simulation studies shows that the proposed realization gives better performance compared to existing realizations in terms of bit error rate (BER) and convergence rate.

High performance architectures for recursive loop algorithms