针对复杂背景下目标边缘检测问题, 研究了分形在目标检测中的应用, 提出了基于分形特征融合的目标边缘检测算法。首先求取图像的分形维数和几何度量空间变换率, 然后运用Dempster-Shafer证据理论对分形维数和几何度量空间变换率进行融合处理, 对图像像素分类, 从而得到目标边缘, 最后细化边缘得到目标检测结果。实验表明, 该算法能够克服复杂背景的干扰, 有效地提取目标边缘。

Aiming at the issue of target edge detection under complex background, we studied the application of fractal in target detection, and put forward a method for edge detecting based on the fusion of fractal features. Firstly, fractal dimension and fractal area measurement of image were calculated out. Then, the two fractal features were fused together using Dempster-Shafer evidence theory. The edge of target was obtained after classifying pixels of image. Finally, the edge was refined and target detection was completed. Experiments demonstrated that the proposed method can overcome the interference of the complex background, and extract the target edge efficiently.