Anushikha Singh

ORCID: 0000-0003-3560-4050
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Biometric Identification and Security
  • Advanced Steganography and Watermarking Techniques
  • Retinal Imaging and Analysis
  • Glaucoma and retinal disorders
  • Chaos-based Image/Signal Encryption
  • COVID-19 diagnosis using AI
  • Digital Imaging for Blood Diseases
  • Spectroscopy and Chemometric Analyses
  • Digital Media Forensic Detection
  • Radiomics and Machine Learning in Medical Imaging
  • Phonocardiography and Auscultation Techniques
  • Lung Cancer Diagnosis and Treatment
  • ECG Monitoring and Analysis
  • Identification and Quantification in Food
  • Handwritten Text Recognition Techniques
  • User Authentication and Security Systems
  • Music and Audio Processing
  • Retinal Diseases and Treatments
  • Potato Plant Research
  • Face recognition and analysis
  • Smart Agriculture and AI
  • EEG and Brain-Computer Interfaces
  • Image Retrieval and Classification Techniques
  • Speech and Audio Processing
  • Image Processing and 3D Reconstruction

Indian Institute of Technology Delhi
2019-2024

Amity University
2013-2018

Amity University
2016-2018

Universidad de Las Palmas de Gran Canaria
2016

Guru Gobind Singh Indraprastha University
2014-2015

Roswell Park Comprehensive Cancer Center
2010

Glaucoma is an eye disorder which caused due to irreversible, progressive damage of optic nerve that leads loss vision. Often in early stage, people are not able realize they affected by glaucoma because there will no symptom like pain or sudden a non-curable disease and hence detection very essential. This paper proposes automated image processing approach for may be diagnostic tool help ophthalmologist mass screening suspects. The proposed based on the segmentation disk cup computing...

10.1109/medcom.2014.7005981 article EN 2014-11-01

This work presents a new approach for automatic recognition of insects through intelligent systems. Insect species employ set sound signals communication purposes which are specie-specific. Based on this fact, an acoustic signal method has been designed to allow efficient taxonomic classification animal group. In paper, the have characterized by Mel and Linear Frequency Cepstral Coefficients (MFCCs LFCCs) compare their efficacy. Then, Support Vector Machine algorithm used achieving average...

10.1109/spin.2016.7566778 article EN 2016-02-01

Late Blight is one of the most common and devastating disease for potato crops in all over world. For less use pesticide to minimize loss crops, identification late blight necessary. The conventional method based on visual assessments which a time consuming process involves manpower. proposed work presents image processing automated from leaf images. In method, adaptive thresholding used segmentation affected area image. threshold value calculated using statistical features makes system...

10.1109/tsp.2017.8076090 article EN 2017-07-01

Segmentation of Optic disc (OD) from a retinal image is essential step while developing automated screening systems for eye disease like diabetic retinopathy, Glaucoma etc. This paper proposes method automatic optic disk segmentation based on region growing technique with seed selection. In this centre considered as to apply segment the preprocessed image. Automatic detection done by double windowing method. The algorithm uses processing techniques contrast adjustment, morphological...

10.1109/ic3i.2014.7019713 article EN 2014-11-01

This paper proposes a method of inserting digital pattern having patient identity in the medical image without tampering information image. To attain imperceptible insertion frequency domain approach is used mid band discrete cosine transform. The original and stego-image compared analyzed for all features also tested retaining information. Blood vessels have been extracted from stego it has established experimental results that remains unaltered. Texture indicates variation texture minimal...

10.1109/tsp.2015.7296372 article EN 2015-07-01

The paper proposes a dynamic thresholding based image processing technique for the detection of hemorrhages in retinal images. algorithm uses information about color and size as tool classifying from other dark lesions present concepts contrast enhancement, background estimation intensity variation at edges that is gradient magnitude supported by some morphological operations. follows simple approach step removal unwanted features targeted images using morphology without compromising with...

10.1109/ic3.2014.6897158 article EN 2014-08-01

The proposed work presents a zero watermarking method for to solve the issue of medical image security telemedicine, tele-radiology & tele-opthalmology applications. This provides tele-medicine application without tempering and no loss clinical information. Local features in Singular value decomposition (SVD) domain are used generate digital binary code (Master Share) each fundus image. master share is strategically combined with encrypted patient ID resulting into secret share. At diagnosis...

10.1109/tsp.2016.7760932 article EN 2016-06-01

Diabetic retinopathy (DR) is a leading cause of blindness in diabetic patients. Exudates are one the most common earliest signs retinopathy. Automatic and accurate detection exudates fundus images an important step early diagnosis DR. In proposed method exudates, two independent approaches based on intensity thresholding morphological processing strategically combined to detect any small present while removing all possible types false positives. This strategic combination removes noise...

10.1109/icumt.2015.7382452 article EN 2015-10-01

Optic disc segmentation is a crucial step in diagnosis of various ocular diseases like Glaucoma and Diabetic Retinopathy. This work proposes technique for automatic detection optic from the fundus images using edge based active contour fitting method. The proposed has used image processing techniques such as smoothing filters removal blood vessels, morphological operations to correctly segment reject false positives, snake model boundary. results obtained are compared with ground truth...

10.1109/ic3.2016.7880227 article EN 2016-08-01

The conventional digital watermarking schemes uses an arbitrary pattern as the watermark which has limitations in proving ownership of watermark. This paper proposes a proficient generation technique from biometric data will be unique and can logically owned to prove ownership. issue is addressed this paper. fingerprint used generate that stamp generated been studied for uniqueness identification images. Discrete cosine transformation embedding image. Experimental results indicate survive...

10.1109/tsp.2013.6614031 article EN 2013-07-01

The existing digital watermarking schemes use a pattern like pseudorandom number sequence, logo image or signature as the watermark. Use of such watermarks is not convincing ownership watermark clear for claims. This paper proposes an efficient generation technique from biometric iris data which will be unique and can owned to prove ownership. issue addressed in this paper. used generate that has stamp generated found clearly identification. Discrete cosine transformation image. Experimental...

10.1109/iccc.2013.6731697 article EN 2013-12-01

This paper proposes a proficient digital watermark generation technique from biometric data which will be unique and can logically owned to prove ownership. The pattern of iris is used generate the that has clear stamp generated been studied for uniqueness identification audio signals. Dither modulation quantization applied on singular values Singular Value Decomposition domain embedding watermark. Experimental results indicates survive signal processing attacks such as Gaussian noise...

10.1109/tsp.2013.6614024 article EN 2013-07-01

This article focuses on the writer verification using safe handwritten passwords smartphones. We extract and select 25 static dynamic biometric features from a character password sequence an android touch-screen device. For we use classification algorithms of WEKA framework. Our 32 test persons wrote generated with length 8 characters. Each person their 12 times. The approach works 384 training samples supervised system. best result 98.72% success rate for correct classification, proposal...

10.1109/ic3.2015.7346651 article EN 2015-08-01

This study presents an approach to use different biomedical signals in order do biometric identification. The are captured using Arduino and Libelium platforms, what offers a low cost solution. used Electromyogram, Electrocardiogram the Galvanic Skin Response. These parametrized well-known measures Neural Networks as classifier develop user result of success rate 85.55% is understood promising way identify people by their signals.

10.1109/spin.2016.7566783 article EN 2016-02-01

In this paper, a bark recognition algorithm for plant classification is presented. A Least-Square Support Vector Machine (LSSVM) with image and data processing techniques used to implement general purpose automated classifier. Using base of 40 sections photographs taken each the 23 species, we applied an homogenize illumination images. After applying it, obtained 256-elements array from Local Binary Pattern (LBP) histogram. Each element was introduced in LSSVM classification. The success...

10.1109/ic3.2016.7880233 article EN 2016-08-01
Coming Soon ...