基于生成对策网络的视频图像动目标识别方法。
Video Image Moving Target Recognition Method Based on Generated Countermeasure Network.
机构信息
School of Information Engineering, Xuzhou University of Technology, Xuzhou 221018, China.
School of Computer Science & Technology, China University of Mining and Technology, Xuzhou 221116, China.
出版信息
Comput Intell Neurosci. 2022 Aug 19;2022:7972845. doi: 10.1155/2022/7972845. eCollection 2022.
In order to improve the accuracy of video image moving target recognition and shorten the recognition time, a video image moving target recognition method based on a generation countermeasure network is proposed. Firstly, the image sensor is used to collect the video image and obtain the video image sequence. The Roberts operator is used for edge detection and Gaussian smoothing of the video image. Secondly, the normalization method is used to extract the key features of moving targets in video images. Finally, training is carried out alternately to generate the countermeasure network model, and the video image moving target recognition sample results are output according to the training results to realize the video image moving target recognition. The experimental results show that the highest recognition accuracy of the proposed method is 98.1%, and the longest recognition time is only 5.7 s, indicating that its recognition effect is good.
为了提高视频图像运动目标识别的准确性和缩短识别时间，提出了一种基于生成对抗网络的视频图像运动目标识别方法。首先，利用图像传感器采集视频图像，得到视频图像序列。然后采用 Roberts 算子对视频图像进行边缘检测和高斯平滑处理。其次，采用归一化方法提取视频图像中运动目标的关键特征。最后，通过交替训练生成对抗网络模型，根据训练结果输出视频图像运动目标识别样本结果，实现视频图像运动目标识别。实验结果表明，所提方法的最高识别准确率为 98.1%，最长识别时间仅为 5.7s，表明其识别效果较好。
Zotero 插件