CAJ | 학술논문

自然图像和计算机图形的鉴别可采用模式识别的方法.采用统计矩特征量和基于颜色滤波阵列的统计量来建立模型,以捕获自然图像和计算机图形在图像内容上的不同相关性.选用哥伦比亚大学自然图像和计算机图形数据库来测试该模型,采用人工神经网络作为分类器进行训练和测试.实验结果表明,该模型的识别率高,稳定性好,具有较好的应用前景.
자연도상화계산궤도형적감별가채용모식식별적방법.채용통계구특정량화기우안색려파진렬적통계량래건립모형,이포획자연도상화계산궤도형재도상내용상적불동상관성.선용가륜비아대학자연도상화계산궤도형수거고래측시해모형,채용인공신경망락작위분류기진행훈련화측시.실험결과표명,해모형적식별솔고,은정성호,구유교호적응용전경.
Pattern recognition method can be used to discriminate photo images from computer graphics. In this paper, the model is established based on statistical moment features and statistics extracted from Color Filter Array(CFA). This model can capture different correlation of image content between photo images and Computer Graphics(CG). To evaluate the performance of the scheme, the model was further tested applying to the Columbia photographic images and photorealistic computer graphics dataset. Artificial neural network is chosen as a classifier to train and test the given images. Experimental results demonstrate that the model performs higher degree of accuracy, indicating that the proposed approach has better stability and possesses promising capability in discrimination of computer graphics from photo images.