http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
서석배,강대성 동아대학교 공과대학부설 생산기술연구소 1998 生産技術硏究所硏究論文集 Vol.3 No.1
A new neural network algorithm and preprocessing filters for handwritten numeral recognition presented in this paper. The perprocessing filter normalizes input numeral pattern and eliminates noise using a human recognition. MLFSOM(Multi-Level Fuzzy Self Organizing Map) proposed in this paper converts SOM(Self Organizing Map) suructure to multi-level for considerate the principal rate and introduces fuzzy algorithm to improve the parameters of SOM. The results of simulstion show that recognition rate in improved and training times are reduced while SOM size reduced.
서석배,강대성,김대진 東亞大學校 附設 情報通信硏究所 1998 情報通信硏究所論文誌 Vol.6 No.1
This paper presents a new vector quantization (VQ) technique using principal component analysis(PCA). Prior to designing a codebook, the proposed approach extracts class features from a image using wavelet transform that has the subband coding characteristics. The computer simulation results show the improved learning time as compared with the SOM(Self-Organizing Map) based VQ.
유전 알고리즘과 개선된 자율 적응 군집화 알고리즘을 이용한 영상 영역화
하성욱,윤호성,서석배,강대성 東亞大學校 附設 情報通信硏究所 1997 情報通信硏究所論文誌 Vol.5 No.1
This paper proposes the new image segmentation method using AACA(advaneced adaptive clustering algorithm) and GA(genetic algorithm). The solution in the general GA is able to be moving because of stochastic reproduction. To cope with these problems, we proposes GBR(great bast reinsertion) and the technique that resolves the premature convergence problem. We also proposes image segmentation method for optimizing vigilance parameter and cluster value. Finally, We uses the new method that removes the small regions in the segmented results. The experimental results show that the proposed algorithm(SGACA; segmentation using genetic adaptive clustering algorithm) gives better performance than the AACA ones.