面向5G无人船的深度图片压缩方法

Deep Image Compression Method for 5G Unmanned Surface Vessels

海上无人船网络信号受天气、信号塔距离等因素影响,无法维持稳定,导致采集的海上船舶图片不能及时回传至服务器;同时海上船舶图片中存在大量的海水、天空等冗余信息,导致关键信息传输速率较低,影响船舶的监控效果。为提高海上无人船的图片传输速率,提出基于语义分割和动态调整码率相结合的无人船深度图片压缩(DCUV)方法。DCUV方法根据图片不同区域的感兴趣程度,在采取不同码率进行图片压缩的同时保留了关键信息。实验结果表明,DCUV方法将海上船舶图片的数据存储空间压缩至原始图片的1.8%时,船舶识别准确率可维持不变。

The network signal of unmanned surface vessels on the sea is affected by factors such as weather and signal tower distance,which cannot maintain stability,resulting in the collected images of ships on the sea not being transmitted back to the server in a timely manner;At the same time,there is a large amount of redundant information such as seawater and sky in the images of ships at sea,resulting in a low transmission rate of key information and affecting the effectiveness of ship monitoring.To improve the image transmission rate of unmanned surface vessels at sea,a deep image compression(DCUV)method based on semantic segmentation and dynamic rate adjustment for unmanned surface vessels is proposed.The DCUV method adopts different bitrates for image compression based on the degree of interest in different regions of the image,while retaining key information.The experimental results show that when the DCUV method compresses the data storage space of marine ship images to 1.8%of the original image,the accuracy of ship recognition can be maintained unchanged.