An Improved Edge Detection Method for Image Analysis in Diverse Domains | ||||
| Aswan Science and Technology Bulletin | ||||
| Volume 2, Issue 2, December 2024, Page 11-29 PDF (1.43 MB) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/astb.2024.310816.1004 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
sara yahya  1; hameda elsanary1; M. Hassan1; abdelmgeid ali2
| ||||
| 1computer science section, mathematics department, faculty of science, Aswan university | ||||
| 2Computer Science Department, Faculty of Computers and Information, Mania University | ||||
| Abstract | ||||
| Edge detection plays a crucial role in image analysis across numerous fields, including medical imaging, industrial inspection, and computer vision. The proposed method has been significantly improved and modified to obtain optimal results in identifying and delineating edges within various types of images. This innovative approach is designed to be universally applicable, transcending specific domain characteristics and proving effective across a wide spectrum of image types and sources. The core of the method leverages gradient information extracted from the image, combined with an improved edge response function. This function is specifically engineered to precisely identify edges with high accuracy and sensitivity. Following the initial edge detection, a series of carefully tuned morphological operations are applied to enhance and refine the detected edges, resulting in clearer and more defined edge representations. Extensive experimental results, conducted on diverse image datasets encompassing multiple domains and applications, demonstrate the method's exceptional effectiveness and versatility. When compared with classical edge detection algorithms such as Canny, Sobel, Prewitt, Roberts, zero crossing, and Laplacian of Gaussian (LOG), the proposed method consistently exhibits superior performance across various metrics and visual assessments. The robust performance and adaptability of this edge detection technique underscore its significant potential for broad adoption across multiple domains in image processing and computer vision. The experiments were conducted with R2015a (MATLAB 8.5) on a machine with an Intel Core i7 processor, 16GB of RAM, and an NVIDIA GTX 1060 GPU, ensuring that the proposed method operates efficiently across different computing environments. | ||||
| Keywords | ||||
| Noise reduction; Image processing; Classical methods; PSNR; MSE | ||||
|
Statistics Article View: 138 PDF Download: 238 |
||||
