Edge detection is a traditional and fundamental task that is regarded as the forerunner of the most widely researched problems in computer vision. In this paper, we present a new robust edge detection method with real-time implementation potential. For edge extraction a 3*3 kernel employed. We obtain differences in intensities at various kernel locations in the suggested edge response function by examining the various 3*3 kernel entrance scenarios to the borders. Each window is divided into two "L"-shaped parts that are rotated before the differences between them are added. The proposed method produces a dense edge response map that can be fed into other methods, such as deep learning architectures. The proposed edge detector was compared to two tried-and-true edge detectors, yielding a compromised result.
