The electrocardiogram (ECG) stands as a cornerstone in the diagnosis of cardiovascular diseases (CVD). ECG serves as a diagnostic tool, allowing healthcare professionals to identify a spectrum of heart conditions. Despite advancements in medical technology, a substantial portion of ECG records persists in paper format, making manual assessment laborious and time-intensive; however, by digitizing these records, automation can be introduced into the diagnosis and analysis processes, streamlining healthcare practices and offering the advantage of easy storage within hospital databases, ensuring accessibility whenever required. Converting ECG images into digital data is a key innovative aspect of this approach. The digitization of ECG images requires a multi-phase approach due to the direct capture nature of the images. Phase I, preprocessing, encompasses tasks such as noise removal, edge detection, and thresholding to prepare the image for further processing. In Phase II, postprocessing techniques including binarization, erosion, and dilation are applied to enhance image clarity and quality, followed by contouring in preparation for analysis. Phase III focuses on classification using ML and DL models to interpret the digitized ECG data. Experiments is conducted to identify the relevance of each phase and select suitable techniques. SVM displayed the highest precision, recall, and F1 score at 0.92, closely followed by XGBoost at 0.89. Thus, SVM emerged as the superior model with an accuracy of 0.915 compared to others.