M. Sridevi

Comparative Study of Image Forgery and Copy-Move Techniques

Image forgery means manipulation of the digital image to conceal some meaningful or useful information of the image. There are cases when it is difficult to identify the edited region from the original image. The detection of a forged image is driven by the need of authenticity and to maintain integrity of the image. This paper surveys different types of image forgeries. The survey has been done on existing techniques for forged image and it also highlights various copy – move detection methods based on their robustness and computational complexity.

Multilevel threshold selection for image segmentation using soft computing techniques

Multilevel thresholding is the method applied to segment the given image into unique sub-regions when the gray value distribution of the pixels is not distinct. The segmentation results are affected by factors such as number of threshold and threshold values. Hence, this paper proposes different methods for determining optimal thresholds using optimization techniques namely GA, PSO and hybrid model. Parallel algorithms are also proposed and implemented for these methods to reduce the execution time. From the experimental results, it is inferred that proposed methods take less time for determining the optimal thresholds when compared with existing methods such as Otsu and Kapur methods.

Parallel Algorithms for Edge Detection in an Image

Edge detection is an important process in image segmentation, object recognition, template matching, etc. It computes gradients in both horizontal and vertical directions of the image at each pixel position to find the image boundaries. The conventional edge detectors take significant time to detect the edges in the image. To reduce the computational time, this paper proposes parallel algorithms for edge detection with Sobel, Prewitt and Robert first order derivatives using a Shared Memory - Single Instruction Multiple Data (SM - SIMD) parallel architecture. From the experimental results, it is inferred that the proposed parallel algorithms for edge detection are faster than the conventional methods.

Optimized Multilevel Threshold Selection Using Evolutionary Computing

Thresholding is the method used for segmenting an image to isolate regions of interest from the image. The result of segmentation mainly depends on the selection of proper threshold values and number of classes. This paper proposes a method for optimal selection of threshold values using Evolutionary computing. The proposed method decomposes the given image to reduce its size so that it can be processed faster using Genetic Algorithm. The resultant image is finally mapped onto the original image space. The efficiency of the proposed method is compared with the other multilevel thresholding techniques namely GA-Otsu and GA-Kapur with and without wavelets. From the experimental results, it is inferred that the proposed method takes less time for processing and provides better results compared to existing methods.

Optimized Channel Allocation Using Genetic Algorithm and Artificial Neural Networks

As the spectrum for wireless transmission gets crowded due to the increase in the users and applications, the efficient use of the spectrum is a major challenge in today’s world. A major affecting factor is the inefficient usage of the frequency bands. Interference in the neighboring cells affects the reuse of the frequency bands. Some of the quality of service parameters such as residual bandwidth, number of users, duration of calls, frequency of calls and priority are considered for optimized channel allocation. Genetic Algorithm and Artificial Neural Networks are applied to determine the optimal channel allocation considering the quality of service parameters. The simulation results show that using Genetic algorithm betters heuristic method and artificial neural networks performs better than Genetic Algorithm by a comfortable margin.

Enhanced Parallel Image Retrieval Technique Using Relevance Feedback

The retrieval of relevant images to a given query is a challenging problem. Many researchers have proposed solutions to solve this problem. The idea of ‘Human in Loop’ adds more useful data to the existing data, which can help refine the output of content-based image retrieval (CBIR) system. In this paper, we extend on ‘Human in Loop’ by setting up a relevance feedback system. It helps the CBIR system to understand closer similarities between images which mere feature vectors and algorithms can’t identify. This paper aims at improving the performance and efficiency of a CBIR system by using Bitmaps to show relevance among the vast number of images, for which feature vectors have already been extracted and stored, and parallel indexing and comparison process are performed to reduce computation time. Use of Bitmap helps learn the feedback obtained from multiple users regarding the relevance between various images until a saturation point is reached. The results show its superiority in comparison with existing models such as QBIC, MARS and VIPER techniques.

Semitransparency Effect in a Video Using Deep Learning Approach

Online learning platforms deliver huge amount of content through video lectures. In academic videos, content written on surface is often obscured by the instructor. This problem is addressed using either pen digitizers or frame averaging method. However, method using pen digitizer requires costly setup while method using frame averages has certain use case limitations. This paper uses deep learning approach to achieve semi transparency effect with better quality and also removes drawbacks of previous works. The experimental results showed that the proposed method works well on different type of videos.

Self-organizing neural networks for image segmentation based on multiphase active contour

AbstractImage segmentation is a process of segregating foreground object from background object in an image. This paper proposes a method to perform image segmentation for the color and textured images with a two-step approach. In the first step, self-organizing neurons based on neural networks are used for clustering the input image, and in the second step, multiphase active contour model is used to get various segments of an image. The contours are initialized in the active contour model with the help of the self-organizing maps obtained as a result of first step. From the results, it is inferred that the proposed method provides better segmentation result for all types of images.n

Optimised template matching techniques in parallel environment with exhaustive search and swarm intelligence

Template Matching (TM) techniques are widely used for recognition and location of objects in image and signal processing applications. The existing techniques such as cross, zero and normalised cross-correlations for TM are time consuming. This paper proposes parallel algorithms for TM problems using Normalised Cross Correlation and Particle Swarm Optimisation (PSO) to suit real-time applications. Experimental results show that parallel version of PSO is comparatively efficient.