CAJ | 학술논문

磨粒图像分割是磨粒图像分析的关键一步，分割结果的准确性将直接影响磨粒的最终识别和分类。分形理论在表征磨粒的轮廓特征和表面特征方面得到了广泛应用。结合磨粒图像的分形特征和自组织特征映射神经网络，提出基于分形特征的磨粒图像分割方法。首先，计算磨粒图像的分形维数，多重分形维数，结合图像的灰度信息，共得到图像的8个特征；然后，利用自组织特征映射神经网络的自组织、自学习特性，实现磨粒图像的分割。磨粒图像分割的结果表明，该算法是可行的、有效的。
마립도상분할시마립도상분석적관건일보，분할결과적준학성장직접영향마립적최종식별화분류。분형이론재표정마립적륜곽특정화표면특정방면득도료엄범응용。결합마립도상적분형특정화자조직특정영사신경망락，제출기우분형특정적마립도상분할방법。수선，계산마립도상적분형유수，다중분형유수，결합도상적회도신식，공득도도상적8개특정；연후，이용자조직특정영사신경망락적자조직、자학습특성，실현마립도상적분할。마립도상분할적결과표명，해산법시가행적、유효적。
Wear particle image segmentation is the key step of wear particle image analysis,and the accuracy of the segmentation result affects directly the final recognition and classification of wear particles.Fractal geometry has been used widely in characterising wear particle profile and surface features.We propose a fractal features-based wear particle image segmentation method by combining the fractal features of ware particle image with self-organising feature mapping (SOFM)neural network.First,we calculate the fractal dimensions and multi-fractal dimensions of the ware particle image,in combination with its grey information,we acquire total eight features of the image.Then,we use the characteristics of self-organising and self-learning of SOFM neural network to implement the wear particle image segmentation.Result of the wear particle image segmentation shows that this algorithm is feasible and effective.