journal6 ›› 2015, Vol. 36 ›› Issue (4): 12-17.DOI: 10.3969/j.cnki.jdxb.2015.04.004

• Information and computer • Previous Articles     Next Articles

Time-Frequency Spectrogram Segmentation Using the Multi-Scale Morphological Gradient and the Marked Watershed Algorithm

 GUO  Hai-Tao, XU  Lei, ZHAO  Hong-Ye, TIAN  Yuan-Yuan, JIAO  Sheng-Xi   

  1.  (1.College of Electronic Information Engineering,Inner Mongolia University,Hohhot 010021,China;2.Electrical Engineering College,Northeast Dianli University,Jilin 132012,Jilin  China;3.School of Automation Engineering,Northeast Dianli University,Jilin 132012,Jilin  China;4.School of Mechanical Engineering,Northeast Dianli University,Jilin 132012,Jilin  China)
  • Online:2015-07-25 Published:2015-07-27

Abstract: It is important and difficult to segment a time-frequency spectrogram due to its strong interference as well as serious overlap among targets and between targets and interference.An image entropy based fusion method for multi-scale morphological gradient image of a time-frequency spectrogram is presented.By combining that method with the marked watershed algorithm,a method for segmenting time-frequency spectrogram based on the multi-scale morphological gradient image and the marked watershed algorithm,is obtained.The experiment results show that it is more practical than the segmentation method which is based on single scale morphological gradient and marker watershed,and more accurate than Otsu method.

Key words: time-frequency spectrogram, multi-scale morphological gradient image, image entropy, watershed algorithm, image segmentation

WeChat e-book chaoxing Mobile QQ