Multipixel image fuzzy edge extraction algorithm based on geometric algebra

Abstract

The traditional image edge extraction algorithm does not integrate the RGB and HSV spatial geometric features effectively, resulting in large errors and long computing time. Therefore, a fuzzy edge extraction algorithm based on geometric algebra for multi-pixel images is proposed and its effectiveness and superiority are verified. The experimental results show that the model has excellent performance in the prediction, the lowest prediction error is 1.21%, the average accuracy is 91.23%, the recall rate is 90.506%, and the F1 value is 90.78%. At the same time, the false positive rate is only 5.48%, showing more accurate error control ability. Overall, the method proposed in this study is of great significance to the field of image edge extraction and can promote the development and progress of related fields.