CAJ | 학술논문

智能化的三维图像分辨技术是三维图像识别技术发展的重要步骤。传统的三维图像分辨技术采用基于模板的识别方法或者基于神经网络的处理方法，无法提取三维图像各个部位的详细信息，从而正确识别率低。提出一种基于三维图像智能分辨在局部酉特征指导下的实现方法，对三维图像进行模块划分，将三维图像详尽的分割为多个小模块，小模块之间实现无缝连接，在模块分割的基础上，提取各个模块之间的酉特征，然后将酉特征进行融合处理，实现整个三维图像的智能化分辨。采用实际的三维图像进行测试实验，结果显示，采用基于局部酉特征指导方法，三维图像正确分辨概率被提高到了98%，具有很好的工程应用价值。
지능화적삼유도상분변기술시삼유도상식별기술발전적중요보취。전통적삼유도상분변기술채용기우모판적식별방법혹자기우신경망락적처리방법，무법제취삼유도상각개부위적상세신식，종이정학식별솔저。제출일충기우삼유도상지능분변재국부유특정지도하적실현방법，대삼유도상진행모괴화분，장삼유도상상진적분할위다개소모괴，소모괴지간실현무봉련접，재모괴분할적기출상，제취각개모괴지간적유특정，연후장유특정진행융합처리，실현정개삼유도상적지능화분변。채용실제적삼유도상진행측시실험，결과현시，채용기우국부유특정지도방법，삼유도상정학분변개솔피제고도료98%，구유흔호적공정응용개치。
According to the traditional method, the template-based method or neural network processing method were used, it cannot consider all parts of the face to extract detailed information on the three dimensional image, so the correct rate of identify was low. The realization of intelligent three dimensional image distinguish under the guidance of local unitary char-acteristic was proposed, firstly, the face was divided into different modules, each small detailed module connected together and it reflected all the three dimensional image , then the local unitary characteristic was extracted out, and the final distin-guishing of entire face was finished. Finally, two face images were used to do experiment, the result shows that with local uni-tary characteristics, the probability of correctly distinguish face image increase to 98%, and it has good value for application.