Particle Swarm Optimisation for Edge Detection in Noisy Images

<p>Detection of continuous and connected edges is very important in many applications, such as detecting oil slicks in remote sensing and detecting cancers in medical images. The detection of such edges is a hard problem particularly in noisy images and most edge detection algorithms suffer from producing broken and thick edges in such images. The main goal of this thesis is to reduce broken edges by proposing an optimisation model and a solution method in order to detect edges in noisy images. This thesis suggests a newapproach in the framework of particle swarm optimisation (PSO) to overcome noise and reduce broken edges through exploring a large area and extracting the global structure of the edges. A fitness function is developed based on the possibility score of a curve being fitted on an edge and the curvature cost of the curve with two constraints. Unlike traditional algorithms, the new method can detect edges with greater continuity in noisy images. Furthermore, a new truncation method within PSO is proposed to truncate the real values of particle positions to integers in order to increase the diversity of the particles. This thesis also proposes a local thresholding technique for the PSObased edge detection algorithm to overcome the problem of detection of edges in noisy images with illuminated areas. The local thresholding technique is proposed based on themain idea of the Sauvola-Pietkinenmethod which is a way of binarisation of illuminated images. It is observed that the new local thresholding can improve the performance of the PSO-based edge detectors in the illuminated noisy images. Since the performance of using static topologies in various applications and in various versions of PSO is different , the performance of six different static topologies (fully connected, ring, star, tree-based, von Neumann and toroidal topologies)within threewell-known versions of PSO (Canonical PSO, Bare Bones PSO and Fully Informed PSO) are also investigated in the PSO-based edge detector. It is found that different topologies have different effects on the accuracy of the PSO-based edge detector. This thesis also proposes a novel dynamic topology called spatial random meaningful topology (SRMT) which is an adoptation version of a gradually increasing directed neighbourhood (GIDN). The new dynamic topology uses spatial meaningful information to compute the neighbourhood probability of each particle to be a neighbour of other particles. It uses this probability to randomly select the neighbours of each particle at each iteration of PSO. The results show that the performance of the proposed method is higher than that of other topologies in noisy images in terms of the localisation accuracy of edge detection.</p>

Particle Swarm Optimisation for Edge Detection in Noisy Images

<p>Detection of continuous and connected edges is very important in many applications, such as detecting oil slicks in remote sensing and detecting cancers in medical images. The detection of such edges is a hard problem particularly in noisy images and most edge detection algorithms suffer from producing broken and thick edges in such images. The main goal of this thesis is to reduce broken edges by proposing an optimisation model and a solution method in order to detect edges in noisy images. This thesis suggests a newapproach in the framework of particle swarm optimisation (PSO) to overcome noise and reduce broken edges through exploring a large area and extracting the global structure of the edges. A fitness function is developed based on the possibility score of a curve being fitted on an edge and the curvature cost of the curve with two constraints. Unlike traditional algorithms, the new method can detect edges with greater continuity in noisy images. Furthermore, a new truncation method within PSO is proposed to truncate the real values of particle positions to integers in order to increase the diversity of the particles. This thesis also proposes a local thresholding technique for the PSObased edge detection algorithm to overcome the problem of detection of edges in noisy images with illuminated areas. The local thresholding technique is proposed based on themain idea of the Sauvola-Pietkinenmethod which is a way of binarisation of illuminated images. It is observed that the new local thresholding can improve the performance of the PSO-based edge detectors in the illuminated noisy images. Since the performance of using static topologies in various applications and in various versions of PSO is different , the performance of six different static topologies (fully connected, ring, star, tree-based, von Neumann and toroidal topologies)within threewell-known versions of PSO (Canonical PSO, Bare Bones PSO and Fully Informed PSO) are also investigated in the PSO-based edge detector. It is found that different topologies have different effects on the accuracy of the PSO-based edge detector. This thesis also proposes a novel dynamic topology called spatial random meaningful topology (SRMT) which is an adoptation version of a gradually increasing directed neighbourhood (GIDN). The new dynamic topology uses spatial meaningful information to compute the neighbourhood probability of each particle to be a neighbour of other particles. It uses this probability to randomly select the neighbours of each particle at each iteration of PSO. The results show that the performance of the proposed method is higher than that of other topologies in noisy images in terms of the localisation accuracy of edge detection.</p>

Object Detection using Particle Swarm Optimisation and Kalman Filter to Track Partially occluded Targets

Motion estimation, object detection, and tracking have been actively pursued by researchers in the field of real time video processing. In the present work, a new algorithm is proposed to automatically detect objects using revised local binary pattern (m-LBP) for object detection. The detected object was tracked and its location estimated using the Kalman filter, whose state covariance matrix was tuned using particle swarm optimisation (PSO). PSO, being a nature inspired algorithm, is a well proven optimization technique. This algorithm was applied to important real-world problems of partially-occluded objects in infrared videos. Algorithm validation was performed by realizing a thermal imager, and this novel algorithm was implemented in it to demonstrate that the proposed algorithm is more efficient and produces better results in motion estimation for partially-occluded objects. It is also shown that track convergence is 56% faster in the PSO-Kalman algorithm than tracking with Kalman-only filter.

Detection of Cyberbullying Through BERT and Weighted Ensemble of Classifiers

Due to the recent popularity and growth of social media platforms such as Facebook and Twitter, cyberbullying is becoming more and more prevalent. The current research on cyberbullying and the NLP techniques being used to classify this kind of online behaviour was initially studied. This paper discusses the experimentation with combined Twitter datasets by Maryland and Cornell universities using different classification approaches like classical machine learning, RNN, CNN, and pretrained transformer-based classifiers. A state of the art (SOTA) solution was achieved by optimising BERTweet on a Onecycle policy with a Decoupled weight decay optimiser (AdamW), improving the previous F1-score by up to 8.4%, resulting in 64.8% macro F1. Particle Swarm Optimisation was later used to optimise the ensemble model. The ensemble developed from the optimised BERTweet model and a collection of models with varying data representations, outperformed the standalone BERTweet model by 0.53% resulting in 65.33% macro F1 for TweetEval dataset and by 0.55% for combined datasets, resulting in 68.1% macro F1.

Detection of Cyberbullying Through BERT and Weighted Ensemble of Classifiers

Due to the recent popularity and growth of social media platforms such as Facebook and Twitter, cyberbullying is becoming more and more prevalent. The current research on cyberbullying and the NLP techniques being used to classify this kind of online behaviour was initially studied. This paper discusses the experimentation with combined Twitter datasets by Maryland and Cornell universities using different classification approaches like classical machine learning, RNN, CNN, and pretrained transformer-based classifiers. A state of the art (SOTA) solution was achieved by optimising BERTweet on a Onecycle policy with a Decoupled weight decay optimiser (AdamW), improving the previous F1-score by up to 8.4%, resulting in 64.8% macro F1. Particle Swarm Optimisation was later used to optimise the ensemble model. The ensemble developed from the optimised BERTweet model and a collection of models with varying data representations, outperformed the standalone BERTweet model by 0.53% resulting in 65.33% macro F1 for TweetEval dataset and by 0.55% for combined datasets, resulting in 68.1% macro F1.

Speech Scrambling based on Particle Swarm Optimisation

Securing any communication system where important data may be transmitted through the channel is a very crucial issue. One of the good solutions in providing security for the speech is to use speech scrambling techniques. The chaotic system used in security has properties that make it a good choice for scrambling speech signal and the optimisation algorithm can provide a perfect performance when used to enhance the hybrid of more than one method. In this paper, we suggest a system that uses an optimisation method, namely, particle swarm optimisation. The evaluation measures prove that the output of the optimisation method has better performance among the methods used in the comparison, including chaotic maps and hybrid chaotic maps.