http://10.9.150.37:8080/dspace//handle/atmiyauni/1282 2026-04-27T19:00:15Z 2026-04-27T19:00:15Z Pithiya, Kiran Ashok Kothari, Ashish M. http://10.9.150.37:8080/dspace//handle/atmiyauni/2293 2025-01-28T10:10:19Z 2024-12-01T00:00:00Z

Title: Image And Video Watermarking: Machine Learning And Frequency Domain Approach For Achieving Dual Security Of Important Information Authors: Pithiya, Kiran Ashok; Kothari, Ashish M. Abstract: Digital watermarking is a technique used to embed a hidden message, such as a company logo, creator’s name, or other identifying information, into a cover medium like images, audio, or video. Effective digital watermarking algorithms must satisfy three key requirements: robustness, perceptibility, and payload capacity. These algorithms are categorized into two primary domains: spatial and transform. In the spatial domain, watermarking modifies pixel values directly based on the watermark. In the transform domain, the frequency components of the cover medium are manipulated to embed the watermark. Lastly, a hybrid approach combines DCT, DWT, and SVD techniques to leverage the strengths of all three methods, achieving improved performance. To evaluate watermarking effectiveness, two metrics are calculated: (1) Perceptibility: This is assessed using Peak Signal-to-Noise Ratio (PSNR) and Mean Square Error (MSE) at the transmitter end. (2) The quality of the recovered watermark at the receiver side is determined by calculating the correlation between the recovered and original watermarks. For assessing robustness, which measures the quality of the watermark retrieved at the receiver side, the correlation between the recovered watermark and the original message is calculated. A higher correlation indicates stronger robustness, ensuring the watermark remains intact even under various attacks or distortions. This thesis introduces a novel method for embedding watermark messages into the crucial part of an image, specifically the region of interest (ROI). The focus is on identifying the human face within an image and embedding the watermark within this detected face. Once the watermark is embedded, the altered face image is reinserted back into the original image. To achieve this, the method employs a face identification algorithm alongside two frequency domain transforms viz. Discrete Cosine and Wavelet as well as the powerful linear algebra technique known as Singular Value Decomposition (SVD). Additionally, the thesis utilizes three widely recognized visual quality metrics to evaluate the method’s effectiveness. At the transmitter side, Peak Signal to Noise Ratio (PSNR) and Mean Square Error (MSE) are used to assess the perceptual quality of the image after watermark embedding. At the receiver side, Correlation is measured to test the robustness of the algorithm. Finally A hybrid algorithm is introduced, which integrates Discrete Cosine Transform (DCT), Discrete Wavelet Transform (DWT), and Singular Value Decomposition (SVD). This combination harnesses the advantages of all three methods, enhancing the overall performance of the watermarking process. Experimental results demonstrate that the hybrid algorithm outperforms individual methods (DCT, DWT, or SVD alone) in both perceptibility and robustness. This makes the proposed approach a superior choice for applications requiring reliable and imperceptible digital watermarking.

2024-12-01T00:00:00Z Sheth, Kinjal Ravi Dr. Vishal, S. Vora http://10.9.150.37:8080/dspace//handle/atmiyauni/1406 2024-03-26T06:57:04Z 2023-12-20T00:00:00Z

Title: An Intelligent Approach to Detect and Classify Facial Image Forgery Authors: Sheth, Kinjal Ravi; Dr. Vishal, S. Vora Abstract: In the era of digital image manipulation and the pervasive presence of social media, the verification of facial image authenticity has become an essential concern. Facial retouching in supporting documents can have adverse effects, undermining the credibility and authenticity of the information presented. In official identification documents, such as passports or driver's licenses, retouching can hinder accurate identification and security measures, potentially leading to identity fraud and security risks. This study presents a comprehensive investigation into the classification of retouched face images using a fine-tuned pre-trained VGG16 & ResNet50 model with ImageNet weight. We explore the impact of different train-test split strategies on the performance of the model and also evaluate the effectiveness of two distinct optimizers namely Adam and RMSprop. The model generalizability has been checked over two standard datasets ND-IIITD retouched faces and MDRF (Multi Demography Retouched Faces- Caucasian samples). The experiment results indicate that the ResNet50 model, fine-tuned with the RMSprop optimizer, attains a maximum accuracy of 98.52% for ND-IIITD and an impressive 99.17% for MDRF (Caucasian). In addition, an examination of various train-test split ratios over these datasets reveals the 80%-20% split ratio as the optimal choice for the approach. Moreover, the experiments show that this method effectively performs on both balanced and imbalanced datasets, emphasizing its robustness and adaptability. In conclusion, the intelligent approach leverages transfer learning and model selection offers a robust solution for the automated detection and classification of facial retouching. This contribution not only enhances image authenticity and trustworthiness in the digital age but also emphasizes the importance of considering various factors, such as model selection and dataset characteristics and hyperparameters in achieving optimal results in this field.

2023-12-20T00:00:00Z Patel, Miralben J. Dr. Ashish, Kothari http://10.9.150.37:8080/dspace//handle/atmiyauni/1405 2024-03-26T06:13:08Z 2023-12-20T00:00:00Z

Title: Develop an Automatic Road Network Extraction System from Remote Sensing Images Authors: Patel, Miralben J.; Dr. Ashish, Kothari Abstract: In recent years, both the development of high resolution satellite images and the amount of newlineavailable aerial images have expanded and accessible easily. Unfortunately, the technologies newlineused to analyze all of these images have not kept pace, so a large portion of the job is still newlinedone manually by humans, which is costly, time-consuming, and error-prone. Due to these newlinefactors, there is a strong need for efficient and dependable techniques that can automatically newlineanalyze remote sensing images. However, even in high-resolution remote sensing images, newlinebackground and roadways might be difficult to distinguish from complex background images newlinebecause of the occlusion of trees and buildings. Now a day, Deep Learning, which has the newlinecomputing power for massive data, has emerged as the most popular and effective newlineclassification approach. newlineTwo different novel methods are developed for automatic road surface detection from high- newlineresolution remote sensing images based on deep learning that detect effectively, efficiently newlineand fast in manner. First, a modified U-Net is used to construct a semantic segmentation newlinealgorithm for road surface extraction. The modified U-Net has fewer convolution layers than newlinethe normal U-Net. The intersection over union (IOU) yielded a model performance of newline93.71%, while the average segmentation time for a single image was 0.28 seconds. newlineThe second proposed approach, Gradient Descent Sea Lion Optimization (GDSLO) fusion of newlineSea Lion Optimization (SLnO) and Stochastic Gradient Descent (SGD) algorithms. The newlinemodel performance for road surface is measured by evaluation metrics, such as precision, newlinerecall, and F1-measure with the highest values of 0.888, 0.930, and 0.810, respectively. The newlineroad edge performance detected with precision, recall and F1-score are 0.801, 0.76, and 0.786 newlinerespectively. In the same way precision, recall, and F1-score for centerline detection is 0.800, newline0.762, and 0.7999 respectively.

2023-12-20T00:00:00Z Vala, Mehulbhai Kuvarbhai Dr. Vishal S, Vora http://10.9.150.37:8080/dspace//handle/atmiyauni/1290 2024-01-25T10:57:23Z 2023-01-31T00:00:00Z

Title: Performance Analysis of Adaptive Data Dissemination in Vehicular Networks Authors: Vala, Mehulbhai Kuvarbhai; Dr. Vishal S, Vora Abstract: Recent advancements in mobile communications, embedded systems, and sen- sors lead to the design of intelligent vehicles. Such vehicles are able to establish wireless communication among themselves, and this is called vehicular ad hoc net- works. A plethora of applications covering safety, e ciency, and infotainment are possible through the use of vehicular networks. Enhancing safety and e ciency is the main goal of an intelligent transportation system (ITS). That is why it is gaining much interest among the research community and automobile industries. Safety-related messages need to propagate e ciently and reliably among mov- ing vehicles to realize safety applications. Variable vehicle density and road topolo- gies in vehicular networks raise many challenges for e cient message dissemina- tion. Furthermore, vehicles must extend message awareness beyond the transmis- sion range of the sending vehicles. The characteristics of vehicular networks, as well as the need to disseminate safety messages over a greater distance, necessitate e cient and reliable multi-hop communications. The current thesis ts into this background and aims to investigate and pro- pose novel and e cient data dissemination protocols, primarily addressing safety applications via vehicle-to-vehicle communication. First, it provides a detailed analysis of message dissemination protocols and their classi cations. The thesis focuses on location-assisted message broadcasting for message dissemination tasks. Native broadcasting methods result in high redundancy and channel contention. Delay-based broadcasting techniques are e cient solutions to reduce excessive re- dundancy and channel congestion. This work provides a comparative analysis of di erent delay-based broadcast techniques. Subsequently, an enhanced adaptive protocol design is presented that is ro- bust against varying vehicle densities and road topologies. The proposed protocol is scalable to accommodate diverse application requirements. Additionally, the behaviour and e ectiveness of the proposed protocol are carefully examined in a realistic environment.

2023-01-31T00:00:00Z