关键词: 图像分割, 离散傅里叶变换, 矩不变法, 阈值化, 傅里叶谱

Abstract: Aiming at the problems of the threshoding method based on the Moment-Preserving Principle (MPP), this paper presented an image thresholding method based on Fourier spectrum and MPP. Firstly the Cumulative Distribution Function Value (CDFV) was calculated according to the MPP to get the rough threshold value, then Discrete Fourier Transform (DFT) was carried out to the histogram of the image to be segmented and compute its Fourier Spectrum (FS), finally the CDFV and the FS were used to change the rough threshold value. Owing to good adaptation of FS,to the rough threshold, this adjustment could get an optimal threshold. Experimental results show that the proposed method can get better segmentation result and better generalization with high speed.

Key words: image segmentation, Discrete Fourier Transform (DFT), moment-preserving principle, thresholding, Fourier spectrum