Edge Detection Using Guided Sobel Image Filtering

Edge detection plays a vital role in numerous engineering and scientific applications, serving as a crucial technique for identifying disruptions, irregularities, boundaries, and other significant features. However, the identification and detection of correct edges are not straightforward, as edge detection depends on image quality parameters such as blur, noise, and edge strength. The traditional edge detection methods, which are based on the gradient of the pixels, suffer from various drawbacks, including false acceptance and rejection of edges. Emerging methodologies, such as those incorporating soft computing or adopting a holistic approach, have been introduced to enhance edge detection. However, these methods often come with increased computational complexity. The accuracy of edge detection heavily depends on factors such as the learning strategy, fitness function, and careful fine-tuning of learning parameters. In this work, we propose a hybrid scheme that considers both weighted guided image filtering and the Sobel mask for accurate edge detection. The weighted guided image filtering enhances edges, while the Sobel mask is used for edge detection. We conducted experiments on a wide variety of images and databases, and it has been found that the proposed mechanism is superior to recently proposed methods.