复杂背景下边缘提取与目标识别方法研究

Study on edge detection and target recognition in complex background

通过对传统形态学边缘提取方法的分析,提出了基于形态学多结构元边缘提取算子,该算子既有良好的边缘提取特性,又很好地解决了噪声抑制和保持图像边缘细节之间的矛盾,通过灰度加权平均值作为阈值进行二值化,更加突出了边缘效果。针对目标成像特点,在提取图像中边缘的像素数、复杂度和最小外接矩形长宽比等多个特征的基础上,通过计算图像中目标边缘的特征评价函数和隶属度函数,利用模糊综合评判技术进行了目标识别。模拟试验表明:基于形态学的多结构元算子具有较强的噪声抑制能力,可以很好地提取复杂背景下的目标边缘;模糊综合评判技术可准确提取目标,较好地解决了复杂背景下的目标识别的难题。

Based on the analysis of traditional edge detection operator of mathematical morphology, a multi-structuring elements edge detection operator of mathematical morphology was proposed, which could suppress noise as much as possible while preserving fine details. The threshold acquired by weighted average of gray levels was used to binarize the image to improve image edge. On the basis of characters of edge pixels, complexity and aspect ratio of minimum enclosing rectangle, a overall fuzzy evaluating technique was utilized to recoginize the target by calculating the characteristic evaluating function and membership degree function. The results of simulation experiments demonstrate that the proposed method could suppress noise effectively and extract target edge from complex background efficiently, and the target in complex background could be detected reliably by overall fuzzy evaluating technique.