摘要： 针对现有计算机视觉理论框架在指导视觉问题中很难克服精度差、受噪声影响大、计算复杂性高的问题，提出仿人的闭环视觉系统模型，引入反馈机制和高层知识的指导，并将其应用到车牌定位系统中。研究结果表明，以该模型实现的仿人视觉车牌定位系统容错性好、准确率高，可以有效地解决光照不均、牌照褪色，以及复杂背景干扰等情况下的车牌定位问题。

关键词: 计算机视觉, 人类视觉, 反馈机制, 车牌定位

Abstract: Aiming at poor accuracy affected by noise and high complexity of calculating in computer vision theoretical framework guiding vision, a new humanoid vision of the closed-loop system model is put forward, a feedback mechanism and the guidance of high-level knowledge is introduced, and they are applied to vehicle license plate location system. Results show that the system not only can accurately locate vehicle license plate and have high fault tolerance, but also can effectively solve location problem under the circumstance of uneven illumination, depigmentation or complex environments interferences.

Key words: computer vision, humanoid vision, feedback mechanism, vehicle license plate location

中图分类号: 

• N945.12