Fast Food Recognition Method Based on Markov Random Field

Abstract

The invention discloses a fast food recognition method based on a Markov random field, which mainly solves the problem of lack of feature points and irregular shape of food in food image recognition, which leads to difficulty in food recognition. The realization process is: (1) establishing a search database , and extract the feature descriptors in the retrieved image to build an index file; (2) extract the feature descriptors of the queried image; (3) find out the similarity of the queried image in the food label class in the search library according to the feature descriptors (4) According to the likelihood score of the queried image in the label food in the retrieval database and the conditional probability between the label categories in the queried person's menu, construct the Markov energy formula, minimize the energy formula, Get the food label class of the queried image. The invention can quickly and accurately identify and classify food images, and can be easily extended to larger retrieval databases and inquired image sets, and can be used to cultivate healthy eating patterns.

Description

technical field [0001] The invention belongs to the field of image information processing, in particular to a food recognition method, which can be used to solve the problems of food image analysis and food category recognition. Background technique [0002] With people's attention to dietary health, it is very necessary for those who want to improve their eating habits to record dietary information in a simple and easy way. So far, 24-hour food records have used the common method of recording "diet history" to assess an individual's eating habits. However, self-reporting methods are often inaccurate, especially among overweight people, some of whom routinely underreport their caloric intake. If the nutritional information of a diet can be automatically detected through food images, it will liberate users from manually recording information. [0003] Recently, in the field of "life-log" research, people are more and more interested in acquiring and processing information a...

AI Technical Summary

Problems solved by technology

However, since food images lack the above-mentioned meaningful feature points, and secondly, since the shape of real food is often amorphous, it is difficult to use the above method to judge the similarity of food shape and cannot accurately classify food

Images