多特征融合下连续帧图像运动目标识别方法  

Moving object recognition method for continuous frame image with multi⁃feature fusion

作　　者：刘朝霞 LIU Zhaoxia(Taiyuan University of Technology,Taiyuan 030024,China)

机构地区：[1]太原理工大学,山西太原030024

出　　处：《现代电子技术》2025年第5期49-52,共4页Modern Electronics Technique

摘　　要：针对不同类型运动目标形状不同、运动模式不一致导致识别难度大的问题,文中提出一种多特征融合下连续帧图像运动目标识别方法。通过摄像机对目标进行连续帧图像采集,从中提取图像颜色及纹理特征,经归一化操作和特征间的欧氏距离融合提取多特征后,进行近似距离计算并排序,将前一部分作为长短期记忆网络输入特征向量,构建运动目标识别模型,输出连续帧图像运动目标识别结果。通过实验验证,该方法能够通过一种可视化的页面将识别结果显示出来,便于用户快速、准确地理解和分析识别结果,识别速度快、稳定可靠,可以为相关领域提供有效的技术支持。Aiming at the difficult recognition caused by different shapes and inconsistent motion modes of different types of motion objects,a continuous frame image motion object recognition method based on multi-feature fusion is proposed.The object is subjected to continuous frame image acquisition by a camera,and the image color and texture features are extracted from the continuous frame image.After the multiple features are extracted by normalization and Euclidean distance fusion among features,the approximate distance is calculated and the sorting is performed.The previous part is used as the input feature vector for the long short-term memory network(LSTM),and a motion object recognition model is constructed to output the result of the motion object recognition for continuous frame image.Experiments verified that the proposed method can display the recognition results with a visual page,which facilitates users to understand and analyze the recognition results quickly and accurately.Its recognition speed is fast,its recognition process is stable and its recognition results is reliable,so it can provide effective technical support for related fields.

关 键 词：多特征融合 连续帧图像 运动目标识别 特征提取 颜色特征 纹理特征 欧氏距离 长短期记忆网络 

分 类 号：TN911.73-34[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]