一种针对煤炭颗粒图像的双阶段自适应分割框架

打开文本图片集

中图分类号：TD67 文献标志码：A

Abstract：In practical production scenarios，the irregular geometric morphologyand complex spatial distribution of coal particles not only affect segmentation accuracy butalso make manual annotation of segmentation masks extremely inconvenient，limiting theirapplicability to large-scale industrial scenarios.To address thisproblem，a dual-stageadaptive segmentation framework （DASeg）forcoal particle imageswas proposed.The framework consisted of the DS-YOLO object detection model，the Adaptive Box Refinement （ABR）module，and the SAM2 image segmentation model.The DS-YOLO model introduced the Dynamic Upsampling （DySample） module and the Spatial and Channel Synergistic Atention （SCSA） module into the neck network of YOLOvl1，which effectively improved object detection accuracy.To solve the problem that the detection boxes generatedbyDS-YOLOdid notclosely fit the actualcoal particle boundaries，theABRmodule was designed.The ABR module performed weighted fusion of the original detection boxes and the bounding boxes of the masks according to weighting coefficients to generate more accurate prompt boxes.The corrected coordinate information was then used as prompt input forthe SAM2 model，which extracted global and local featuresand fused prompt region information to generate target masks，therebyachieving coal particle segmentation.Experimental resultsshowed that the DASeg framework performed excellently in coal particle image segmentation tasks， with a Pixel Accuracy （PA） of 93.1% ，a Mean Intersection Over Union （mIoU） of 88.4% ， and a Mean Dice （mDice） of 93.4%

Key words： coal particle image segmentation; object detection; YOLOv11； Adaptive Box Refinement; image segmentation; SAM2

0引言

近年来计算机视觉发展迅速，基于图像分割的粒度检测方法因其高效、精确和直观的特点，逐渐成为研究和应用的热点[]。（剩余11933字）