This paper addresses the problems encountered during digitization and preservation of inscriptions such as perspective distortion and minimal distinction between foreground and background. In general inscriptions possess neither standard size and shape nor colour difference between the foreground and background. Hence the existing methods like variance based extraction and Fast ICA based analysis fail to extract text from these inscription images. Natural gradient flexible ICA (NGFICA) is a suitable method for separating signals from a mixture of highly correlated signals, as it minimizes the dependency among the signals by considering the slope of the signal at each point. We propose an NGFICA based enhancement of inscription images. The proposed method improves word and character recognition accuracies of the OCR system by 65.3% (from 10.1% to 75.4%) and 54.3% (from 32.4% to 86.7%), respectively.

No abstract

Coronavirus is a contagious disease that affects individuals in a large scale. Coronavirus had a huge impact on the nation's economy and human lifestyle. The motivation behind this study was establishing a better diagnosis test for coronavirus infection. The RT-PCR test is used to diagnose the coronavirus frequently and returned a negative result for an infected individual. Furthermore, this test remains prohibitively expensive for most citizens, and not everyone could afford it due to financial hardship. An efficient imaging approach is de veloped for the evaluation of lung conditions, which has been done by examining the chest X-ray or chest CT of an infected person. Deep Learning is the well-suited sub domain of Artificial Intelligence [AI] technology, which offers helpful examination to consider more number of chest X-rays images that can basically have an effect on coronavirus screening. The goal of this research is to cluster the radiograph images present in the dataset into COVID-19, healthy and viral pneumonia by making use of the artificial neural networks. The training dataset was fine-tuned with eleven previously trained convolutional neural architectures. The assessment of the models on a test sample shows that AlexNet, DenseNet-121, GoogleNet and S queezenet1.1 as the top performing models.

No abstract

No abstract