基于深度哈希与注意力机制的花卉图像检索

Flower Image Retrieval Based on Depth Hash and Attention Mechanism

针对当前的花卉识别方法在真实场景下容易受背景、光照等因素干扰导致识别准确率低、识别速度慢的问题,提出一种基于深度哈希与注意力机制相结合的图像检索方法用于花卉识别。上述方法在神经网络中融合了注意力机制用于降低背景干扰提升特征质量,并增加一个哈希层降低特征维度以提升检索效率,在图像预处理阶段采用自适应直方图均衡化降低光照干扰影响。实验结果表明,在更接近真实场景的自制花卉数据集True Flowers上,所提方法与传统神经网络方法相比平均检索精度提升了1.3%,检索速度提升了156倍,在公共数据集Oxford 17 Flowers上新方法的准确率要高于其它文献方法,由此证明了新方法的有效性和先进性。

Aiming at the problems of low recognition accuracy and slow recognition speed caused by the interference of background,illumination,and other factors in the actual scene,an image retrieval method based on depth hash and attention mechanism for flower recognition is proposed.This method integrates the attention mechanism in the neural network to reduce the background interference and improve the feature quality,adds a hash layer to reduce the feature dimension to improve the retrieval efficiency,and uses adaptive histogram equalization to reduce the influence of illumination interference in the image preprocessing stage.The experimental results show that on the self-made flower data set True Flowers closer to the actual scene,this method's average retrieval accuracy and retrieval speed are improved by 1.3%and 156 times compared with the traditional neural network method.On the public data set Oxford 17 Flowers,the accuracy of this method is higher than that of other literature methods,which proves the effectiveness and advancement of this method.