A Fridge Food Management System Based On Image Processing Technique

With the development of electronic information technology, smart home technology enter into human life, the smart fridge as an important part of the smart home, attracted the attention of many research institutes and enterprises. Looking at the variety of smart refrigerator concept, its main features include:multimedia functions, that a smart refrigerator can play audio, images, video, cable, and display the electronic document; food management functions, that a intelligent refrigerator can identify not only the type of food bust also the quantity information, and if the food is not enough, the refrigerator will remind its users to buy, furthermore the refrigerator can recommend recipes to user, advocate the healthy eating; network connectivity, that a smart refrigerator can be connected via phone or Internet with the user, so that users can remotely see the food in the refrigerator, and to buy food selectively. There is no doubt that the food management function is the core of the intelligent refrigerator.Smart refrigerator food management function, this paper presents the solution of the food management system based on image processing technology. Of the system by access to food image stitching, image segmentation and extraction of food, food images classification and identify three main modules.(1) Food image acquisition and mosaic. In this paper, a set of cameras is placed to obtain the food image in the refrigerator, limited by the camera parameters; each camera just can get a local image of the refrigerator. Then image mosaic technology is used to obtain a whole food images from these local images. Firstly to correct image distortion by image rectification approach, then using the coordinate transformation and image fusion method to combine the local image.(2) Food image segmentation and extraction. Because there are a variety of foods in the fridge food image, in order to make the food image recognition successful, we need to use image segmentation processing to extract the individual food images. A feature-based image segmentation method is used, firstly extracting the image’s color and texture features, secondly using principal component analysis algorithm to reduce the dimension of image feature vector set, finally using the k-means algorithm to complete the image segmentation.(3) The classification and identification of food images. Different fruits and vegetables have huge differences in color and texture, so in this paper the image classification and recognition method is based on image’s color features and texture features. Firstly according to color characteristics, to separate the difference color image, secondly for the same color images, using texture features to classify.The experimental results show that using the image processing method, this paper initially achieved the food management system’s basic functions.