基于深度多特征融合的自适应CNN图像分类算法

Adaptive CNN-based Image Classification Algorithm Based on Deep Fusion of Multi-Feature

为更好地提取图像内容信息,提高图像分类精度,提出一种自适应卷积神经网络(CNN)图像分类算法。通过融合图像的主颜色特征,利用CNN提取空间位置特征,且针对多特征融合权重值的设定问题,运用改进的差分演化算法优化各特征权值,提高固定权值分类精确度。实验结果表明,该算法分类精度相比CNN算法提升了9.2个百分点,在图像分类中具有较好的分类效果。

To improve the accuracy of image information extraction and image classification,an adaptive Convolutional Neural Network(CNN)-based algorithm is proposed for image classification.The algorithm effectively integrates the main color features of the image,and the spatial position features are extracted by using CNN.For the setting of the weight value of multi-features fusion,an improved differential evolution algorithm is presented to optimize the feature weight,so the accuracy of classification using fixed weight is improved.Experimental results show that the algorithm provides excellent image classification results,and its classification accuracy is 9.2%higher than that of CNN.