Stark H, Barker R C, Lee D
Appl Opt. 1972 Nov 1;11(11):2540-50. doi: 10.1364/AO.11.002540.
An extension of a new technique that enables parallel, simultaneous processing of remotely obtained images is presented. The technique holds out promise for automatic on-board classification of data. The central feature involves the generation of binary masks, directly from the image, based on object reflectance data, that groups objects into equivalence classes. These masks, called equivalence class masks, can be used in various logical combinations to isolate classes of objects with a priori known reflectance or radiance signatures. Experimental verification of the technique is furnished for simple scenes. A computational scheme, based on a sequence of integrated irradiance measurements on the image, that enables the identification of objects within an equivalence class is suggested.
本文介绍了一种新技术的扩展，该技术能够对远程获取的图像进行并行、同步处理。这项技术有望实现数据的自动机载分类。其核心特性包括基于目标反射率数据直接从图像生成二值掩膜，将目标分组为等价类。这些掩膜称为等价类掩膜，可用于各种逻辑组合，以分离具有先验已知反射率或辐射特征的目标类别。文中提供了针对简单场景的该技术实验验证。还提出了一种基于图像上一系列积分辐照度测量的计算方案，用于识别等价类中的目标。
