基于二阶方向导数尺度空间的小目标检测

目标的尺度信息，是2维图像中小目标检测性能发挥的重要因素。该文提出一种基于二阶方向导数尺度空间的小目标检测方法，直接利用目标尺度信息对所感兴趣的目标进行选择。在Lindeberg尺度空间理论的基础上，该文采用偏微分方程方法，推导了最大和最小二阶方向导数的尺度空间，并分析了其在目标尺度表示上的特点。同时，给出构造尺度空间的参数选择依据，分析不同目标在尺度空间上的变化规律，提出利用二阶方向导数尺度空间进行小目标检测的具体实现算法。通过方法对比和对实际数据的处理，表明了该文方法具有较为稳健的小目标检测性能，提高了Laplace尺度空间对非圆结构目标的检测能力。

Small Target Detection Based on Second Order Directional Derivative Scale-space

The information of targets scale is an important factor for small target detection in two-dimensional image. This paper presents a small target detection method based on second order directional derivative scale-space, chooses directly the interest target with the information of targets scale. On the basis of Lindebergs scale-space theory, the maximum and minimum second order directional derivative scale-spaces are derived, with the method of partial differential equations, and the characteristics of targets scale in the scale-spaces are analyzed. Meantime, the selection of parameters for constructing scale-spaces is given. On the analysis of the response of different targets in the scale-spaces, the implement algorithm using the second order directional derivative scale-space to detect small target is proposed. Experiments on the actual image show that the proposed method has a more robust detection performance than the compared one, and the detection capability of Laplace scale-space for non-circular structure target is improved.