CAJ | 학술논문

对非刚性目标的跟踪算法是机器视觉领域研究的热点之一。在线半监督或无监督的机器学分类器算法，因为正负样本选取不准确，从而使跟踪产生漂移现象。本文针对跟踪目标产生的漂移问题，提出了一种基于 Hough 模型的像素级特征描述符，同时，采用基于目标和背景分割的分割模型进行最大概率目标分割，获得了良好的跟踪性能，这种算法能准确提取目标的正确位置。在视频序列上的测试结果证明了本算法有良好的抗目标形状和外观的非刚性变化性能。
대비강성목표적근종산법시궤기시각영역연구적열점지일。재선반감독혹무감독적궤기학분류기산법，인위정부양본선취불준학，종이사근종산생표이현상。본문침대근종목표산생적표이문제，제출료일충기우 Hough 모형적상소급특정묘술부，동시，채용기우목표화배경분할적분할모형진행최대개솔목표분할，획득료량호적근종성능，저충산법능준학제취목표적정학위치。재시빈서렬상적측시결과증명료본산법유량호적항목표형상화외관적비강성변화성능。
Tracking non-rigid target has obtained much attention in the field of machine vision.in online semi-supervised or unsupervised machine learning classifier algorithm,the positive and negative sample selection is not accurate,so the tracker drift.In this paper,we put forward a novel algorithm based on Hough model.,Segmentation model is also used in this paper to obtain a more excellent tracking performance.This algorithm can accurately extract the correct position of the target.Experiment results in several challenging video sequences demonstrate that this algorithm has a excellent appearance and shape non-rigid change resistance.