摘要
针对双目立体视觉系统中弱纹理对象立体匹配困难问题,提出了一种约束随机编码图像生成算法,用于增加弱纹理对象表面的纹理质量。首先,根据立体匹配算法的搜索策略及代价聚合的计算方法,设计了投影图像像素编码规则,建立了投影与目标混合图像位置像素局部唯一性约束下的计算方法,形成高强度的纹理分布图像;然后,将该算法生成的约束随机编码图像投影到航空行李表面,采用立体匹配算法完成行李表面的三维重建,并对重建结果进行评估。实验结果表明,算法可对纯色或纹理特征不明显的航空行李进行准确可靠的三维建模。
Aiming at the difficulties during stereo matching of weak-texture-object in binocular stereo vision system,a constrained random coding image generation algorithm is proposed for enhancement of texture quality.Firstly,according to the search strategy of stereo matching algorithm and calculation method of cost aggregation,the pixel coding rules of the projected image are designed,and the calculation method under local uniqueness constraint of projection and target hybrid image position pixels is established to form a high-intensity texture distribution image.The constrained random coded image generated by the algorithm is projected onto the surface of air baggage,the 3D reconstruction of air baggage surface is performed by stereo matching algorithm,and the reconstruction result is evaluated.The reconstruction experimental results of standard parts and actual baggage show that the proposed algorithm can accurately and reliably model the air baggage with solid color or weak texture features.
作者
高庆吉
彭传波
方榕
GAO Qingji;PENG Chuanbo;FANG Rong(College of Electronic Information and Automation,CAUC,Tianjin 300300,China)
出处
《中国民航大学学报》
CAS
2020年第1期38-42,共5页
Journal of Civil Aviation University of China
基金
天津市自然科学基金项目(12JCZDJC34200)
天津市科技计划项目(14ZCZDCX00055)。
关键词
弱纹理对象
约束随机编码算法
高强度纹理分布
立体匹配
三维重建
自助行李托运系统
weak-texture-object
constrained random coding algorithm
high intensity texture distribution
stereo matching
3D reconstruction
self check-in system with baggage
