由于回转窑烧结图像境界模糊、噪声严重,传统方法难以有效地分割出烧结图像中的物料区、火焰区、充分燃烧区和黑把子区等关心区域(ROIs)。提出了利用敏感区域改进双快速行进法与利用能量衰减方程除去区域间耦合的图象分割新方法。首先,在快速行进法中引入了敏感区域的概念,定义了新的终止条件,利用改进后的双快速行进法与图像融合方法进行物料区和黑把子区的粗分割;然后, 提出了采用能量衰减方程除去烧结物料对火焰区的影响,利用大津方法对火焰区进行粗分割,并通过粗分割的火焰区与黑把子区的差分来进行火焰区的精分割。对回转窑烧结图像的分割试验表明,利用本方法能够有效地从回转窑烧结图像精确地分割出ROIs。
It's difficult for traditional methods to segment some Regions of Interesting (ROIs) of material, blackbar, flame and sufficient burning from sintering images owing to their characteristics of week boundary and strong noise. A new image segment method was proposed based on improved dual fast marching method by sensitive region and energy descendent equation to decrease the coupling of regions. Firstly, in the fast marching method, the concept of sensitive region was introduced and a new end condition was defined, and the regions of material and blackbar were segmented by the improved dual fast marching method along with the image combination methods. Secondly, the energy descendent equation was applied to reduce the influence of material region on flame region. The flame region was roughly segmented by ostu method and then was accurately segmented by combining the roughly segmented flame region with blackbar region. The segment experiments show that the new developed methods are valid and accurate to segment the ROIs from the sintering images in rotary kiln.