Digital image processing plays an important role in extracting visual information from natural objects, including the morphological structure of leaves. One of the crucial techniques in this process is edge detection, which is used to emphasize object boundaries and support pattern analysis. This study aims to compare three edge detection methods, namely Sobel, Prewitt, and Canny in recognizing patterns in leaf images using the Python programming language and the OpenCV library. The methods used include quantitative experiments by implementing the three edge detection techniques on grayscale leaf images, followed by visual result analysis based on the criteria of edge sharpness, morphological detail, noise, and computational efficiency. The results show that the Canny method produces the most accurate and clean edge detection, with the ability to capture small details and reduce noise significantly. Sobel shows quite good performance in highlighting the main structure of the leaf, while Prewitt produces simpler and less precise results. Based on the evaluation results, the Canny method is considered the most effective for the purposes of digital leaf pattern classification and analysis. This study provides an important contribution in selecting the optimal edge detection method for computer vision applications in the field of digital botany..