由于红外热像仪采集的热红外影像存在大量的噪声、边缘信息模糊难提取, 因此, 提出了一种基于 Laplace算子和灰色关联度相结合的边缘检测方法。该方法首先引进数学形态学对含有噪声的热红外影像进行形态学滤波, 然后利用 8邻域的 Laplace算子作为参考序列, 计算参考序列与系统序列之间的关联度得出关联度矩阵, 再给定阈值判断是否为边缘点, 最后细化影像。基于 MATLAB进行仿真实验, 结果表明: 该算法对具有噪声的热红外影像具有较好的抗噪效果, 能检测出传统方法不能提取的边缘细节信息, 通过调整阈值可以控制检测的边缘信息量, 是一种较好的边缘检测方法。

There are so much noise in the infrared thermal images collected by imager that it’s difficult to extract edge. Therefore a method combining Laplace operator and gray relation is proposed in this paper. Mathematical morphology is introduced to the process of noise for infrared thermal image. Correlation matrix is obtained by calculating the correlation degree between reference sequence and system sequence, 8 neighborhood of Laplace operator as the reference sequence. The threshold is given to determine whether the pixel is edge point. At last, the image is thinned. Based on MATLAB, the result shows that the algorithm not only has the ability to remove noise, but also can extract the edge which other traditional methods can’t extract. Edge information can be controlled by threshold adjustment and which is an effective edge extraction method.