现代电子技术2025，Vol.48Issue(5)：49-52,4.DOI:10.16652/j.issn.1004-373x.2025.05.008
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
摘要
Abstract
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Key words
multi-feature fusion/continuous frame image/moving object recognition/feature extraction/color feature/texture feature/Euclidean distance/LSTM分类
电子信息工程引用本文复制引用
刘朝霞..多特征融合下连续帧图像运动目标识别方法[J].现代电子技术,2025,48(5):49-52,4.
