摘要
在分析分形理论在目标检测领域应用的基础上,提出了一种基于有向分维参数的海面运动目标自动检测算法.该算法与传统的分形参数计算相比,具有检测准确、计算快速和易于并行处理的特点.根据图像纹理的方向特性,选取相应方向的邻域像素来计算有向分维参数,根据其数值的不同来确定目标的位置,具体实现步骤是,首先利用传统边缘检测方法得到海面运动舰船的航迹位置,再沿着运动航迹方向计算有向分维参数,对所得图像二值化处理,最终通过分析二值图像获得目标位置.该算法适用于具有方向特性的自然背景下运动目标的检测,实验表明能够有效检测出图像中目标所在的位置.
With the analysis application of fractal theory in the field of target detection, this paper proposes an algorithm for automatic detection of moving sea target based on directional fractal dimension. Compared with the traditional fractal dimension, this algorithm has the following properties: higher detection rate, lower computational cost and better parallelizability. According to the directional features of textures in images, we compute the directional fractal parameter with the gray values of the selected pixels of the neighborhood in the corresponding direction. By the difference of directional fractal parameters between the target and background, the position of the target is determined. The detailed procedure involves: Firstly, we obtain the roughly approximate track of the moving ship by the traditional edge detection method. Secondly, we calculate the corresponding directional fractal parameter along the track direction and get the binary image by thresholding. Lastly, the position of the target can be decided by analysis of the above binary image. This algorithm is suitable for the detection of moving target which possess different directional feature from the natural background. Experiments show the effectiveness of this algorithm in the detection of moving targets.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2004年第4期486-490,共5页
Pattern Recognition and Artificial Intelligence
关键词
分形
有向分维
图像处理
目标检测
Fractal
Directional Fractal Dimension
Image Processing
Target Detection