Abstract: In automatic measurement of the radio spectrum parameters, the traditional threshold-based method is difficult to obtain satisfactory result when the spectrum has undulating background noise. To solve this problem, this paper proposed a signal noise separation algorithm based on morphological gradient to separate signals from background noise by detecting edges of signals. The experimental results show that this algorithm can deal with the spectrum with undulating background noise, and get more accurate separation result than the traditional method.

Key words: background noise, morphology gradient, edge detection, signal noise separation, spectrum

摘要： 在频谱参数自动测量中，传统的基于阈值的信号噪声分离方法对于具有不平坦背景噪声的频谱难以取得满意的分离结果。针对这一问题，提出了一种基于形态学梯度的信号噪声分离算法，通过检测信号的边缘实现信号噪声分离。实验结果证明，该算法能够较好地处理背景噪声起伏不平的频谱，分离结果与传统方法相比有较大的改善。

关键词: 背景噪声, 形态学梯度, 边缘检测, 信号噪声分离, 频谱