Select flotation froth image structural element utilizing highlighting information

Aiming at the structural element selection problem in mineral flotation froth image processing, a structural element (SE) selecting method based on bubble highlighting information in this paper. On the basis of binary image granulometries, the morphological pattern spectra theory is introduced. The high lighting components are filtered by a connected filter operator. Meanwhile, Zernike moment is introduced to represent the froth high lighting area. Thus, the size and shape distribution pattern spectra matrix is yielded for connected components, which makes the process do not need a SE. In term of pattern spectra matrix, the mean and deviation pattern spectra value is defined, which is calculated to yield a optimal SE. Experimental results show that the proposed method can utilize flotation froth image information effectively, and an optimal SE can be contributed to froth image segmentation process, which improved the image segmentation accuracy.