The COVID-19 pandemic has a significant impact on the world, putting the health and lives of millions of people in danger. The difficulty in diagnosing COVID-19 at an early stage is a challenging task because many patients could not exhibit any symptoms . The pathogenic laboratory test like RT-PCR are inaccurate they can produce falsenegative results, which increases the risk of human life. The chest X-ray is most easy to capture and low cost radiological images used for detecting COVID-19. In this paper the Convolutional Neural Networks (CNNs) approach is applied over a deep learning system to extract features and detect COVID-19 from chest X-ray images. The datasets used in experiments consists of 3000 chest X-ray images , view images of 1000 normal, 1000 pneumonia, 1000 COVID-19 diseases from the various sources. In this paper the CXR images are resize, gaussian noise without clipping is added to the images for making the proposed model to be more robust and to improve ability for generalizing new data, denoising with DnCNN_sigma25 is used to remove potentially clean images in the hidden layer. Then deep learning feature extraction techniques with Local Binary Pattern (LBP) and Gray Level Co- Occurrence Matrix (GLCM)are carried out for extracting the structural and textual characteristics of the chest X-ray images. This study evaluates the performance of trained model on unseen data and testing set is used to correctly identify the COVID-19 positive and negative cases from CXR images. The pretrained model is compare with other classification models to achieve the better performance parameters in COVID-19 detection. The results of proposed model shows the accuracy, F1 Score, Positive Predicted Value (PPV), Specificity, Sensitivity of 99%, 98%, 97% , 97%, 96% among all other classifier models in COVID-19 detection