CAJ | 학술논문

针对木材缺陷图像，采用基于小生境和交叉算子的粒子群算法（NCSPSO）优化支持向量机（SVM）进行图像分割，提取木材缺陷分割图。主要对NCSPSO算法进行改进，寻找最优惩罚系数C和高斯核函数中的参数，然后采用SVM方法对训练样本进行综合训练，以建立最佳分类模型，并对木材缺陷图像分割测试。与模拟退火法（SA）及原NCSPSO算法进行对比实验，从而验证改进NCSPSO算法的优越性。
침대목재결함도상，채용기우소생경화교차산자적입자군산법（NCSPSO）우화지지향량궤（SVM）진행도상분할，제취목재결함분할도。주요대NCSPSO산법진행개진，심조최우징벌계수C화고사핵함수중적삼수，연후채용SVM방법대훈련양본진행종합훈련，이건립최가분류모형，병대목재결함도상분할측시。여모의퇴화법（SA）급원NCSPSO산법진행대비실험，종이험증개진NCSPSO산법적우월성。
In view of wood defect images,Particle Swarm Optimization(NCSPSO)algorithm based on microhabitat and crossover operator are adopted to optimize support vector machine(SVM)for image segmentation and wood defect segmenta-tion diagram extraction. The focus is placed on the improvement of NCSPSO to seek the optimal penalty coefficient C and the parameter of Gauss kernel function. Then SVM is used to establish the best classification model and test wood defect image segmentation. Simulate anneal arithmetic and the NCSPSO are compared through experiments to verity the superiority of NC-SPSO.