發(fā)布時間：2018-05-02 18:11

  本文選題：車標(biāo)定位 + 散熱器柵格精確分類　； 參考：《計(jì)算機(jī)工程與應(yīng)用》2017年02期

【摘要】：由于車標(biāo)的背景散熱器柵格形狀大小不一、顏色不定、背景多樣,因此導(dǎo)致了車標(biāo)定位的困難,故精確分類散熱器柵格是準(zhǔn)確定位車標(biāo)的基礎(chǔ)。提出了一種基于散熱器柵格背景精確分類的車標(biāo)定位方法,首先依照車牌與車標(biāo)空間位置關(guān)系確定車標(biāo)粗定位,然后依據(jù)柵格紋理特征,利用霍夫變換和灰度值的梯度變化確定散熱器柵格背景的類別,進(jìn)而通過不同算子分別對不同種類柵格背景進(jìn)行背景消融;為了保證多種光照條件下的準(zhǔn)確定位,引入離散度,并將其與大津法進(jìn)行融合,形成一種適用于車標(biāo)定位的自適應(yīng)二值化方法,同時結(jié)合形態(tài)學(xué)對柵格背景進(jìn)一步處理,得到準(zhǔn)確的車標(biāo)定位。這種方法適用于在不同光照強(qiáng)度和不同類型的車標(biāo)背景條件下,對車標(biāo)進(jìn)行定位。對10類車標(biāo)、30類散熱器柵格共1 200張圖像進(jìn)行車標(biāo)定位,實(shí)驗(yàn)結(jié)果顯示,圖像總體的車標(biāo)定位準(zhǔn)確率可以達(dá)到97.67%。
[Abstract]:Because the background radiator grid is different in shape, color and background, it is difficult to locate the vehicle logo, so the accurate classification of radiator grid is the basis of accurately locating vehicle logo. In this paper, a method of vehicle mark location based on radiator grid background classification is proposed. Firstly, the rough location of vehicle mark is determined according to the relationship between license plate and vehicle logo, and then based on the texture feature of grid. The background of radiator grid background is determined by using Hough transform and gradient change of gray value, and the background of different grid background is ablated by different operators. In order to ensure accurate location under various illumination conditions, the dispersion degree is introduced. An adaptive binarization method is formed by combining it with the Otsu method. At the same time, the raster background is further processed in combination with morphology, and an accurate vehicle mark location is obtained. This method can be used to locate vehicle signs under different illumination intensity and different background conditions. A total of 1,200 images of 10 kinds of vehicle marks and 30 kinds of radiator grids were used to locate the vehicle marks. The experimental results show that the accuracy of the overall image can reach 97.67%.
【作者單位】： 中山大學(xué)工學(xué)院智能交通研究中心;廣東省智能交通系統(tǒng)重點(diǎn)實(shí)驗(yàn)室;視頻圖像智能分析與應(yīng)用技術(shù)公安部重點(diǎn)實(shí)驗(yàn)室;
【基金】：國家科技支撐計(jì)劃(No.2014BAG01B04)
【分類號】：U495;TP391.41
，

本文編號：1834954