基于改進(jìn)二維Otsu算法的車道線分割研究

打開文本圖片集

關(guān)鍵詞：車道線分割；二維Otsu;二維直方圖；二值圖像

中圖分類號：TP391 文獻(xiàn)標(biāo)志碼：A 文章編號：1003-5168（2025）08-0034-08

DOI：10.19968/j.cnki.hnkj.1003-5168.2025.08.006

Abstract：[Purposes] The traditional lane line image segmentation algorithm has low recognition accuracy and is susceptible to noise interference， which affcts the performance of the lane departure warning system.Based on this，an optimal oblique dividing line threshold algorithm based on θ - divisionisproposed.[Methods] Firstly，the image collcted bythe vehicle camera is preprocessed by region of interest extraction，grayscale，filtering and denoising to obtain the image that is conducive to lane line segmentation.Secondly，combined with the two-dimensional Otsu algorithm，the two-dimensional histogram is segmented by using a threshold segmentation line rotating around the threshold point ，so that the twodimensional histogram is divided into two parts：the targetand the background，and the inter-class variance between the target and the background is calculated during the rotation of the threshold segmentation line，and the optimal threshold segmentation line is determined when it reaches the maximum.In this way，the boundary and noise region information in the 2D histogram is included in the segmentation decision.[Findings] The experimental results show that the improved algorithm improves the accuracy of lane line segmentation by and the anti-noise performance by . Even in the face of complex road scenes，the algorithm still performs well in lane segmentation.[Conclusions] The proposed algorithm can provide more reliable lane segmentation image for the folow-up lane detection，which is of great significance for the realization of lane departure warning system.

Keywords： lane line segmentation; two-dimensional Otsu; two-dimensional histogram; binary image

0 引言

當(dāng)前，智能交通視頻監(jiān)控系統(tǒng)在交通管理、行政執(zhí)法、刑事偵查和運輸安全等領(lǐng)域中發(fā)揮著越來越重要的作用。（剩余6263字）