Optimizing Image Edge Detection: Fusing Sobel and Robert via Convolutional Neural Networks for Enhanced Performance

image edge detection is a significant area of research as its applications are covered in various engineering and sciences applications. The traditional masking-based edge detection methods are not very accurate as they suffer from false edge acceptance and rejection. The traditional Sobel operator detect edges in horizontal and vertical directions, while Robert operator detect edges in diagonal directions. In this work, a novel method which combine the Sobel and Robert operators edge detection process through convolutional neural networks (CNNs) is proposed. The complete process of the weight optimization is discussed in the paper. The experiment is performed on the Berkley segmentation database. The simulation is performed on MATLAB. The simulation is performed on the wide variety of the images. The results are compared with traditional and state-of-art methods and results are found to the superior to compared methods.