基于YOLO-v8的选通图像目标检测方法研究  

Research on Target Detection Method of Gated Image Based on YOLO-v8

作　　者：张正[1] 付荣珂 田青[1] 

机构地区：[1]北方工业大学信息学院,北京100144

出　　处：《工业控制计算机》2025年第1期76-78,81,共4页Industrial Control Computer

基　　金：国家重点研发计划资助(2020YFB1600702)。

摘　　要：激光距离成像技术具有全天候成像、抗干扰能力强等优点,基于选通图像的目标检测有很大的前景。针对选通图像分辨率低与噪声干扰的问题,采用自适应双直方图均衡化与GDIP模块结合对选通图像进行前处理,提高图像质量;针对选通灰度图像特征不足的问题,基于YOLO-v8网络,改进了neck层,同时增加了改进的注意力机制模块,更好地提取特征信息。在自制数据集上进行验证,实验结果表明该方法能够有效提升检测精度。Laser range imaging technology has the advantages of all-weather imaging and strong anti-interference ability.Target detection based on gated images has great prospects.In order to solve the problems of low resolution and noise interference of the gated image,dual-platform histogram equalization is used in combination with Gaussian filtering and GDIP module to pre-process the gated image to improve the image quality.In order to solve the problem of insufficient features of the gated grayscale image,based on the YOLO-v8 network,the neck layer is improved,and an improved attention mechanism module is added to better extract feature information.Verified on a self-made data set,experimental results show that this method can effectively improve detection accuracy.

关 键 词：选通图像 图像增强 目标检测 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]