CAJ | 학술논문

针对传统自适应阈值法在水下图像分割过程中存在目标与背景粘连等问题，提出一种基于灰度化权值调整的水下彩色图像分割方法。区别于传统阈值分割方法中灰度化权值固定且研究如何确定合适阈值的思路，研究灰度化权值调整方法。本文方法利用统计得到的图像灰度信息来确定不同通道内目标与背景间灰度级差异性的大小，提高差异性较大通道的比例，以增强灰度化后图像中目标与背景的对比度，同时减小差异性较小通道的比例，以降低其对灰度化结果的干扰，对灰度化后的图像进行阈值分割得到最终分割结果。针对水下作业的时序列图像，设计了两步序列图像处理方式来降低耗时。通过4组水下对比实验以及耗时实验对方法的有效性进行了验证。
침대전통자괄응역치법재수하도상분할과정중존재목표여배경점련등문제，제출일충기우회도화권치조정적수하채색도상분할방법。구별우전통역치분할방법중회도화권치고정차연구여하학정합괄역치적사로，연구회도화권치조정방법。본문방법이용통계득도적도상회도신식래학정불동통도내목표여배경간회도급차이성적대소，제고차이성교대통도적비례，이증강회도화후도상중목표여배경적대비도，동시감소차이성교소통도적비례，이강저기대회도화결과적간우，대회도화후적도상진행역치분할득도최종분할결과。침대수하작업적시서렬도상，설계료량보서렬도상처리방식래강저모시。통과4조수하대비실험이급모시실험대방법적유효성진행료험증。
This paper proposes an underwater color image segmentation method based on weight adjustment for col?or?to?gray. The method is proposed in order to solve the inter?regional connection problem in underwater image seg?mentation when using the traditional adaptive threshold method. This method is different from the traditional thresh?old segmentation methods in which the weight of color for gray processing is fixed and it needs to research how to determine appropriate threshold. The proposed method uses the gray level information acquired from the statistical calculation to identify the differences between the targets and the backgrounds in the different image channels. Next, it increases the weight of the channel with higher difference to enhance the contrast in the gray image. Mean?while, this method also decreases the weight of the channel with lower difference to reduce its interference in the gray color processing. The final segmentation result is achieved by threshold segmentation of the image obtained af?ter grey color processing. In regards to the sequence images capture during underwater operation, a two?step se?quence image processing method is designed to reduce the time consumption. The validity of the method is verified through four comparative underwater experiments and one time consumption experiment.