-
摘要: 为回避多尺度边缘检测中融合多幅图像涉及的启发性规则, 同时克服 Jeong 算法的局限性, 本文首先基于 Tan 的广义边缘定义设定一个适应度函数, 然后采用自适应遗传算法针对单个像素及其邻域结构的特征搜索最佳滤波尺度得到局部最佳边缘检测尺度, 最后由最佳尺度高斯滤波器得到最终边缘图像. 实验证明能够得到令人满意的结果.Abstract: To avoid the heuristic rule in multi-scale edge detection and overcome the shortcomings of Jeong's adaptive algorithm, a fitness function based on Tan's generalized edge cost function is proposed in this paper. Adaptive genetic algorithm incorporated with division selecting strategy has been used to seek a set of optimal scales according to each site of the image. By convolving input image with optimal scale Gaussian filters, the final edge map can be obtained at last. Experiments show that the results of adaptive scale filtering algorithm are satisfying.
-
Key words:
- Edge detection /
- multi-scale /
- Gaussian filter /
- adaptive genetic algorithm
-
计量
- 文章访问数: 2629
- HTML全文浏览量: 108
- PDF下载量: 1658
- 被引次数: 0
下载:
