面向智能回收装置的废旧手机深度森林识别

Deep Forest Identification of Used Mobile Phone for Intelligent Recycling Equipment

手机更新速度的日益加快使得面向废旧手机(UMP)的智能回收装置研制成为典型城市再生资源回收再利用领域的研究热点,其首先需要解决的难题是如何基于有限图像样本实现UMP的精准识别。针对上述问题,提出了面向智能回收装置的UMP深度森林识别模型。首先,对UMP图像进行预处理以消除背景信息;接着,提取多尺度梯度直方图特征以获得有效信息;最后,将通过多粒度扫描获取高维特征作为级联森林的输入,并基于自适应网络深度构建非神经网络模式的UMP识别模型。基于电信设备认证中心的典型数据集和回收装置拍摄图像数据集验证了所提方法的有效性。

The increasing speed of cell phone renewal makes the development of intelligent recycling equipment for used cell phones(UMP)a research hotspot in the field of typical urban recycling and reuse,and the first challenge to be solved is how to achieve accurate recognition of UMP based on limited image samples.To address the above problem,a UMP deep forest recognition model for intelligent recycling equipment is proposed.Firstly,the UMP image is pre-processed to eliminate the background information.Secondly,the multi-scale gradient histogram features are extracted to obtain the effective information.Finally,the high-dimensional features are obtained by multi-grain scanning as the input of the cascade forest,and the non-neural network model of UMP recognition model is constructed based on the adaptive network depth.The effectiveness of the proposed method is verified based on a typical dataset of telecommunication equipment certification center and a dataset of images taken by recovery equipment.