Ochotona Curzoniae Image Segmentation Based on Improved CV Model

Hai-yan CHEN, Yu-hui ZENG, Ying LIN

Abstract


Ochotona curzoniae image possessed the characteristics of low contrast, intensity inhomogeneity and complex background. This paper proposed an improved Chan-Vese (CV) model. This model is combined image mean filtering information with object fitting term. As CV model has a problem with easily falling to segment image with object’s intensity inhomogeneity. In addition, to suppress the effects of background interference by using rectangular dirac delta function, to reduce the calculation of initialization of the evolving level set function by using Quadtree, to automatically obtain the initial contour by using Otsu. Tests and comparisons show that the new model has a better effect for Ochotona curzoniae image segmentation in autumn and winter season.

Keywords


Ochotona curzoniae, Chan-Vese (CV) model, Mean filtering, Rectangular dirac delta function, Quadtree, Otsu


DOI
10.12783/dtetr/amma2017/13367

Full Text:

PDF

Refbacks

  • There are currently no refbacks.