Canny edge detection towards deep learning Arabic document classification

The paper describes the implementation of deep learning-based edge detection in image processing. A set of points in an image at which image brightness changes formally or sharply is called edge detection. Using edge detection filters, we can extract the feature of an object. In our work, we aim to develop a deep learning system to classify Arabic document images into four classes as follows: printed, handwritten, historical, and signboard and applying edge detection filters to extract features from document images. We will be using two edge detection methods namely Sobel, and Canny edge detection that are applied in 1000 Arabic document images to extract edges. Analyzing the performance factors are done in the terms of accuracy on the premise of Mean Squared Error (MSE) and python is employed for edge detection implementation. The experimental results show that the Canny edge detection technique results higher than the Sobel edge detection technique.