A Study on Accuracy Recognition of Images using Elimination of Specular Reflection Light

－Trial of Restoration from Peripheral Pixel Information using Spatial Filtering－

Abstract

As the performance of computer systems improves, image recognition technology expands its scope of application and is being implemented in society in a variety of fields. For example, it is one of the important applications to recognize the information about the product from the image of the product displayed in the retail store. However, when recognizing a product and extracting the necessary information, the specular reflection of light from the surface of the package caused by light sources such as ceiling lights acts as noise, reducing the accuracy of computer image recognition. Therefore, the specular reflection can be removed simply by applying a spatial filter to a single image, in order to improve the accuracy of image recognition. In the article, we describe the results of an experiment that improve the accuracy of image recognition.