Abstract:

The success of edge detection operators is very important for computer vision algorithms. A well-drawn edge map will play a vital role in the performance of all subsequent operations. In this study, various city photographs were taken, and a standard set of preprocessing was performed. These obtained images were processed by selected Canny, Sobel and Prewitt edge detection algorithms and compared in terms of performance using real reference value data. In these comparisons, the success of the edge detection algorithms was evaluated by the methods of precision, sensitivity, and f-score calculations. In our literature review, we did not encounter any studies using these criteria in evaluating the performance of edge detection operators. In this study, we evaluated the performances of the Canny, Sobel and Prewitt edge detection operators with these three criteria. Programs and code fragments used for image processing and comparison of algorithms are included in this study. According to the results obtained, no significant superiority could be detected between the Canny, Sobel, or Prewitt edge detection operators on the data set used against the real reference value, but it has been proven that the comparison process can be converted into an objective output. In the second part of this study, other algorithms will be added to the comparison and the study will be expanded and a complete performance test will be obtained.

Keywords: