基于形态学的复杂背景目标检测算法的研究

复杂背景下目标检测存在诸多困难，主要为背景对目标检测的干扰，大量噪声存在导致传统导数边缘检测方法的失效等。针对上述两点，提出了分割区域图像、利用形态学方法检测目标的新算法。首先利用目标与背景灰度差异性来确定目标的大致区域，将其分割出来，然后再结合多结构元素法进行目标的精确检测。通过与原图像分割、聚类算法分割实验比较，该算法具有较好的抗干扰性和抗噪性能。

Target Detection from the Complex Background Based on Mathematical Morphology

There are many difficulties in target detection if the background is complex. The major problem is detection result has been interfered by the background and noise, leading to traditional edge detection methods invalid. Aiming at these problems above, this paper presents a new algorithm, which detectes target using regional images segment combined with mot phologic methods. In detail, the approximate target area is determined and segmented by using the difference of gray between the object and its background, and then the target is detected accurately by multi-structure element of mathematical morphology. Compared with the traditional image segment ways and clustering segmentation algorithm, this algorithm has preferable performance of anti interference and anti-noise in examples.