Colorectal polyp classification (CPC) is essential for the early diagnosis and prevention of colorectal cancer (CC). However, existing methods often struggle to accurately classify polyps due to variations in size, shape, texture, illuminations and imaging artifacts in colonoscopy images. To address these challenges, a novel deep learning based COPN-NET is proposed for CPC. First, the input colonoscopy images are denoised to remove image noise and retain the crucial structural information while preserving the boundaries of the polyps using an Adaptive Wiener Star Filter (AWSF). To enhance the diversity of the data set and the model's ability to generalize, data augmentation methods are used like flipping, rotating, scaling, or adjusting the brightness of the data. The improved images then fed into the lightweight feature extraction ability of MobileNet and the powerful representation learning ability of RegNet together to form a MobileRegNet architecture. The extracted deep features are fed to a Deep Belief Network (DBN) for classification into Normal, Adenoma, and Cancer. The proposed framework enhances the feature representation and classification results. Experimental results show that COPN-NET is able to obtain an accuracy (AC) of 99.08%, precision(PR) of 96.64%, recall (RE) of 95.71%, F1-score (F1) of 96.17%,and specificity (SP) of 95.61%.