COVID-19 is a highly infective viral disease and it is observed that the newest strains of the SARS-CoV-2 virus has greater infectivity rate. Due to the present pandemic, the economy of the country, the mental and physical state of the people and their regular lives are being affected. Medical studies have shown that the lungs of the patients who are infected by the corona virus are mostly being affected. Chest x-ray or radiography is observed to be one of the most effective imaging techniques for diagnosing problems which are related to the lungs. The study proposes a novel COV-XDCNN model with external filter for diagnosing diseases such as COVID-19, Viral Pneumonia, automatically which can assist the healthcare workers, mainly during the time of outbreak. The motivation of this research lies in designing an automated system which can aid the healthcare workers. The proposed model with external filter gives 97.86% test accuracy in classifying the chest radiography images. The model performance is examined with various other models such as NASNetMobile, ResNet50, MobileNet, VGG-16 etc. and analyzed. The model proposed in this study shows better performance than most of the existing traditional methods.