Effect of Illumination and Distance on Identification and Classification of Food Objects Images

Abstract

In this paper we have studied the effect of varying illumination conditions and varying distance of image acquisition on identification and classification of images of food objects like Bonda, Idli, Puri, Somasa and Wada. A backpropagation neural network (BPNN) is developed using 18 color and 24 texture features.The image samples are collected under different lighting conditions like natural lighting, florescent lamp, incandescent bulb and varying acquisition distances like 20, 40, 60, 80, 100 centimeter. The maximum classification accuracy found witht he images taken under natural lighting conditions. The work finds application in autometic food serving by robots in hotels, restaurents, shapping malls, pharmacentical industry, foodindustry, etc.

Keywords

• Food object
• Color features
• Textural features
• Back propagation neural network
• Illumination