CAJ | 학술논문

视频交通参数检测中的车辆分割需要准确地检测与车辆连在一起的阴影和车灯产生的亮斑.一般地,路面与车辆的图像在灰度结构上存在显著差异.本文推导了图像在光照变化情况下的一种灰度结构--极点与极性分布图,并提出了一个基于该分布图的车辆阴影与亮斑检测算法.该算法能够精确地检测车辆阴影,车灯照射产生的路面亮斑和因其他原因被误为车辆的路面.该方法使用对光照变化不敏感的灰度结构,阴影检测准确而稳定,计算量小.
시빈교통삼수검측중적차량분할수요준학지검측여차량련재일기적음영화차등산생적량반.일반지,로면여차량적도상재회도결구상존재현저차이.본문추도료도상재광조변화정황하적일충회도결구--겁점여겁성분포도,병제출료일개기우해분포도적차량음영여량반검측산법.해산법능구정학지검측차량음영,차등조사산생적로면량반화인기타원인피오위차량적로면.해방법사용대광조변화불민감적회도결구,음영검측준학이은정,계산량소.
The shadows similar to the vehicle and the spots caused by vehicle lamps need to be accurately detected in the vehicle segmentation involved in the video-based traffic parameter measurement. Generally, the road surface is different from the vehicle surface in the gray-level architecture. An invariant gray-level architecture-the extremum image in the changing illumination environment is derived and a novel algorithm is presented for detecting shadows and spots. The gray-level structure that is not sensitive to the illumination is employed in the algorithm and the road surface mistaken as vehicles can be removed.