Optimization of the Canny Edge Detection Method Using Optimization Algorithms in Order to Determine the Boundaries of Agricultural Lands

Accurate determination of land boundaries is a vital step in identifying land use and planning for its management. For cropland, this mapping allows farmers and agribusinesses to better estimate land area for effective use of agricultural inputs, such as seeds, pesticides, fertilizers and other resources, and to optimize production and post-production activities. With these interpretations, identifying the precise edges of a crop field determines the success of a program developed using computer vision and machine vision tools. According to the experiments, the Canny detector performs better than other traditional detectors, therefore, in this article, it has been tried to present new methods to improve the performance of the Canny detector by using two meta-heuristic algorithms of ant colony and thermal simulator. To evaluate the proposed methods, we have used three evaluation criteria: MSE, PSNR, and SSIM. The results show that the meta-heuristic algorithms used have made the Canny detector more optimal. When we used the ant colony algorithm, the edges were shown well, and when we used the simulated annealing algorithm, in addition to the edges, the image texture was also preserved.