An Ice masonry image segmentation method based on multi-scale fusion and improved Grabcut

Currently, most of the ice masonry used in ice construction is cut by automated production lines, but its quality inspection is still done manually, which cannot form a completely automated and intelligent system. The reason for this phenomenon is the absence of suitable image-processing algorithms for ice masonry. To address the above issues, this letter proposes a fully automatic image segmentation method based on multi-scale fusion and an improved Grabcut, which is proposed and applied to production lines for ice masonry. This method significantly mitigates the effect of the natural environment on ice masonry images by enhancing the details through multi-scale fusion. The algorithm proposed in this letter automatically segments the ice masonry image using an improved interaction-free Grabcut algorithm with specific rules to precisely select the segmentation region. It is worth mentioning that the method proposed in this letter is the first to be uesd for image segmentation of naturally occurring river ice masonry. It enables fully automated segmentation of objects with semi-transparent and reflective characteristics in production lines. Meanwhile, the segmentation method proposed in this letter allows Grabcut to satisfy the segmentation requirements in just one iteration, which speeds up the processing speed of the algorithm.