基于BBO-MLP和纹理特征的图像分类算法

An image classification algorithm based on BBO-MLP and texture features

为了提高图像分类的准确率,解决多层感知器(MLP)收敛速度缓慢等问题,提出了一种基于生物地理学优化-MLP(BBO-MLP)和纹理特征的图像分类算法。首先,从图像库中选取3类不同的图片,对图像分类算法运行环境进行建模;其次,选取角二阶矩(UNI)、熵(CON)、惯性矩(ENT)和相关性(CDR)4个纹理参数构建一个四维特征矩,根据用户提供的类别号和图像的纹理特征向量生成训练样本文件;然后,将提取的数据作为MLP的输入数据,为MLP定义一个评估栖息地的误差适应度函数并对适应度函数进行全局优化,利用BBO算法训练MLP,得到分类模型;最后,利用训练好的MLP对图像进行分类,并引入二次反馈机制进一步提高算法性能。实验结果表明,与PSO、GA、ACO、ES和PBIL等优化算法相比,本文的BBO-MLP算法具有较高的分类正确率。

In order to improve the accuracy of image classification and the conve rgence speed of multi-layer perceptron （MLP）,this paper presents a novel image classification algorithm based on biogeography-based opt imization （BBO） MLP and texture fe atures.The proposed algorithm steps are as follows.Firstly,three kinds of different images from the image da tabase are selected,and the operating environment of the image classification algorithm is modeled.Secondly,4texture parameters,including UNI,ENT, CON and COR,are selected to obtain 4-dimens ion feature moments,and the training sample files are generated according to the category number provided by the customer and image texture feature vector.Thirdly,defining the evaluation of the habitat error as the fitne ss function,the data are used to train MLPs using BBO algorithm,and the classification model is obtained. Finally,the trained MLPs are used to image classification.Further,feedback mechanism is introduced to improve the performance.The experimental results show that,comparing with the current heuristic learning alg orithms of PSO,GA,ACO, ES and PBIL algorithms,the proposed BBO-MLP method is more effective and feasib le,which has higher classification precision compared with other approaches.