SURYA PRASAD POTNURU
@mvgrce.com
Associate Professor, ECE Department
MVGR COLLGE OF ENGINEERING (AUTONOMOUS)
EDUCATION
DECE,B.TECH,M.TECH,PH.D
RESEARCH, TEACHING, or OTHER INTERESTS
Computer Vision and Pattern Recognition, Electrical and Electronic Engineering, Signal Processing, Artificial Intelligence
Scopus Publications
Scopus Publications
P. Surya Prasad and B. Prabhakara Rao
Springer Singapore
K.V.Ravi Teja, P.Santhosh Kumar, N.Shanmukha Rao, and P.Surya Prasad
IEEE
A new Cascading Algorithm for de-noising images corrupted by Salt and Pepper noise at even high density level has been proposed. The first stage employs a Decision Based Median Filter (DBMF) which is mainly for eliminating the noise from the image, the pixels affected by noise are replaced with the median value of the processing window. The second stage employs, a Decision based Unsymmetric Partial Trimmed Variant Filter (DBUPTVF). It inspects the output of the first stage and eliminates the un-removed noisy pixels in the image based on the number of pixels in the processing window which are noisy, by Partial Trimming. The second stage is mainly used for removing any un-removed noise from the first stage and to enhance the image quality. The proposed algorithm is developed, such that noise of any density level could be removed from the image. The proposed cascading structure exhibits improved noise elimination capabilities than the existing standard filters and other cascading algorithms for up to noise densities as high as 90%. The Proposed de noising Cascading algorithm has been examined for different grayscale images. The Proposed Cascading Algorithm for de-noising, produces lesser Mean Square Error(MSE), better Peak Signal to Noise Ratio(PSNR), improved Image Enhancement Factor(IEF) and higher Structural Similarity Index Metric(SSIM) than several existing algorithms.
Potnuru Surya Prasad and Bhima Prabhakara Rao
Institution of Engineering and Technology (IET)
With the increasing awareness on the reliable distribution of power with good quality, the research in power distribution automation surveillance system has gained prominence. The performance of distribution system is affected significantly by the damaged insulators in numerous ways. With enormous growth in the power distribution network, the traditional methods of examining the lines by manual patrolling and pole climbing to check that in close proximity are not feasible. The blooming field of on-line condition monitoring of electrical equipment aims at predicting the possible failures before they actually occur. The improvement of a proficient, alternative method to assess the condition of insulators in a power distribution system using image processing and machine learning techniques is found to be a satisfactory method. This study presents a system to automatically monitor the insulator of overhead power distribution lines using extraction of features from wavelet transform as well as local binary pattern histogram Fourier (LBP-HF) of the insulators and then subsequent condition analysis done by using support vector machine (SVM). The efficacy of the proposed techniques is validated by the results contained in this study and is found to be suitable for real-time overhead power distribution system monitoring automation.
P. Surya Prasad and B. Prabhakara Rao
IEEE
With ever-increasing awareness on quality and reliable power distribution, the research in the area of automation of distribution system has great relevance from the practical point of view. Electric power utilities throughout the world are more and more adopting computer aided control, monitoring and management of electric power distribution system to offer improved services to the consumers of electricity. The purpose of on-line condition monitoring of cables or any electrical equipment is to predict possible failures before they actually occur. With phenomenal growth of distribution network even to remote areas, the traditional methods of inspecting the lines by foot-patrolling and pole-climbing to check them in close proximity do not seem to be viable. Since the damaged insulators of the distribution system affects the performance of distribution system significantly in terms of reduction in voltage, aerial patrolling has been adopted in developed countries for the purpose of insulator monitoring. The development of an efficient and alternative method for insulator condition monitoring uses image processing and machine learning techniques and is found to be a sustainable method. This work covers automatic defect detection and classification of insulator systems of electric power lines using vision-based techniques.
P. Surya Prasad, , B. Prabhakara Rao, and
International Hellenic University
In the present world, there is a great necessity to have reliable and quality power distribution, and so there is great scope for research on automation of distribution system. The main objective of this paper is to analyze and comprehend different machine learning and image processing based algorithms to find a practical solution for automated inspection of overhead power line insulators. This method is a relatively new approach for. This paper also highlights the constraints and limitations that are present in the various existing methodologies to achieve the objective. Traditionally the workers who inspect these lines check them in close proximity by going for foot-patrolling and pole-climbing. With an incredible expansion of power distribution network even to remote areas, previously mentioned methods do not seem to be viable. The development of an efficient method of condition monitoring by using image processing followed by machine learning techniques is found to be a suitable method and thus emerging as a feasible option for real-time implementation. The few techniques like artificial neural networks (ANN), Hidden Markov Model (HMM), k-means clustering, Wavelet transform features, S-transform features, and support vector machines (SVM) applied in the domain of condition monitoring of the insulators were presented.
D. Hari Hara Santosh, Prasanth Dasari, N. L. S. Kiran, and P. Surya Prasad
IEEE
In this paper different Image compression techniques particularly for texture images are discussed. A texture image can be conceived as repetition of different patterns and spatially the pixels are connected in predefined shapes. So the better way of compressing these texture images is by suitable image compression techniques. Both Finite Radon Transform (FRAT) and Discrete Cosine Transform (DCT) are well suited for this. Applying the DCT on the Radon domain gives the new transformation which is Finite Radon Cosine Transformation (FRCT). The properties of both DCT and FRAT are been exploited in this new transformation. Image quality is measured objectively, using peak signal-to-noise ratio (PSNR). A comparison with a Wavelet transform, DCT and Finite Ridgelet Transform (FRIT) based compression systems are given. The results show that by applying the DCT on the Radon domain (FRAT) one gets high PSNR compared to other transforms for a given compression ratio.
D. Hari Hara Santosh, V. N. Lakshmana Kumar, D. Raja Ramesh, and P. Surya Prasad
IEEE
In this paper, different types of speckle noise and impulse noise removal techniques are presented. Dual Tree complex wavelet transform because of having time and frequency resolution (MRA) and also of its shift invariance property works well in removing the speckle noise. Because of adaptive nature of mask size depending on the noise quantity in the image, adaptive median filter works better in removing the salt and pepper noise. To show the performance of DT-CWT speckle filters like median filter, Lee filter, Frost and Kuan filters and DWT are considered. Adaptive median filter is compared with other filters and also the transformations. The superiority of Dual Tree Complex Wavelet Transform in removing the Speckle noise and Adaptive Median filter in removing the Speckle noise in images are presented. Graphs are drawn between the input PSNR and output PSNR both for speckle noise removal techniques and also impulse noise removal techniques.
V. Sreerama Murthy, P. Surya Prasad, Sumit Gupta, and D.K. Mohanta
IEEE