Identification of products on shop-racks by morphological preprocessing and feature-based detection

Abstract

In this paper we describe the method allowing the identification of products on the shop-racks. This method consists of two steps: morphological preprocessing where the image of the rack is segmented in order to find the split-lines between products; and recognition step where, based on the segmentation results, the feature-point approach is used to identify the products. In the proposed method, thanks to novel preprocessing step, we reduce the search-space to possible locations obtained as results of segmentation.