The global outbreak of the coronavirus, commonly known as COVID-19 has rapidly spread worldwide, affecting nearly every nation and causing significant fatalities. With over 440 million confirmed cases and more than 6.09 million deaths, the need for efficient and cost-effective monitoring of affected individuals is paramount. Chest X-rays are readily accessible, economical, and reliable diagnostic technology. This study proposes a framework utilising restricted Boltzmann machines (RBM) as a feature extractor, categorising extracted features into three classes: COVID-19 positive, normal, and pneumonia. To enhance the model’s performance, we leverage image enhancement techniques, including gamma correction with two gamma values (0.5 and 1.5) and contrast limited adaptive histogram equalization (CLAHE) with two clipLimit values (0.2 and 3.0). The proposed model undergoes rigorous evaluation across various setups and is compared against three existing methods using multiple publicly available datasets. Notably, the proposed model, employing a gamma value of 1.5, achieves outstanding results with an accuracy of 98.21%.