CAJ | 학술논문

受多方面因素的影响,图像在特征空间中的分布是非常不均匀的,往往围绕多个中心.为了解决多个特征中心的问题,提出了一种基于aiNet人工免疫网络的遥感图像检索算法.该算法根据免疫网络机理及相关反馈技术,利用aiNet人工免疫网络对用户的反馈信息进行学习记忆,能有效寻找多个最优解,提高了系统对用户语义的理解能力.由于该网络具有减少冗余、多样性、学习和记忆的特性,避免了传统算法容易陷入局部最优的缺点.实验结果表明,该算法能有效理解用户的反馈信息,提高了传统检索方法的准确性.
수다방면인소적영향,도상재특정공간중적분포시비상불균균적,왕왕위요다개중심.위료해결다개특정중심적문제,제출료일충기우aiNet인공면역망락적요감도상검색산법.해산법근거면역망락궤리급상관반궤기술,이용aiNet인공면역망락대용호적반궤신식진행학습기억,능유효심조다개최우해,제고료계통대용호어의적리해능력.유우해망락구유감소용여、다양성、학습화기억적특성,피면료전통산법용역함입국부최우적결점.실험결과표명,해산법능유효리해용호적반궤신식,제고료전통검색방법적준학성.
Images distribute very uneven in the feature space,sometimes encircle several centers influencing by several factors.A novel remote sensing image retrieval method based on aiNet model of Artificial Immune System(MS) is proposed to settle this mul-ti-center problem.aiNet is used to learn and memorize the user feedback information,so to find multi-global optimum solutions efficiently and improve the recognition of customers' semantic target for system.This network has the properties of reducing re-dundancy,diversity,learning and memorizing,since then the shortcoming of running into local optimum solution of traditional algo-rithm can be avoided.Experimental results show that this approach can recognize the user feedback information efficiently and im-prove the retrieval accuracy.