A HYBRIDIZED MODEL USING INDEPENDENT COMPONENT ANALYSIS WITH AN IMPROVED GENETIC ALGORITHM FOR CLASSIFYING BLURRED TEXT DETECTION IN WILD SCENE

Removing text from images is a challenging task, particularly when dealing with fuzzy borders, shading, and embossing. This process is essential but can be difficult to achieve using existing solutions, which may not be adequate to address the multifaceted issues that occur in the wild. The history of writing spans several centuries and having accurate and rich semantic information in text is crucial in various vision-based use cases. As a result, researchers in the fields of computer vision and pattern recognition have been actively investigating text recognition in natural settings. Deep learning has gained popularity in recent years, resulting in the development of unique, useful, and successful approaches. In this study, we propose a robust approach for detecting blurred text in wild scenes using a hybridized model. Our model integrates independent component analysis (ICA) with an improved genetic algorithm (IGA) to enhance feature extraction and selection, respectively, resulting in improved classification accuracy. We compare our approach with existing state-of-the-art methods using a publicly available dataset and demonstrate that our hybridized model outperforms them, achieving a significantly higher accuracy of around 98.65%. We employ support vector machine (SVM) and k-nearest neighbor (KNN) classifiers to classify the extracted features. The results show that both classifiers perform well, with the KNN classifier achieving slightly higher accuracy than the SVM classifier. This approach can be extended to other image classification tasks in various domains, providing a powerful tool for real-world applications. The findings of this study have significant implications for the development of more accurate and efficient methods for detecting blurred text in challenging environments, paving the way for the use of semantic information in various vision-based applications.