The Design And Implement Of Image Retrival System Based On Content

With the continuous development of electronic consumer products and the growing popularity,more and more of their products have the function of image storage and display capabilities.At the same time,how to effectively manage a huge number and variety of pictures has become an important research topic.The image recognition feature can help users achieve the classification management of the image,facilitate users to quickly retrieve a large number of image you want subject or content,it has great value and practical significance.Based on the current content-based image retrieval research status,this parper chooses the effective image retrieval method to design and implement image retrieval system.This paper presents an image retrieval system based on content.The whole system uses Z-turn Board development board as the hardware platform and the dual-core Cortex-A9 ARM as a processing platform.It completed image processing algorithms and Linux system operation.Furthermore,it provides visual interactive interface in the QT application development.The whole system is divided into two modules according to function design and realization,applications and HCI module respectively.Application cont:rol mainly includes a storage module,a pre-processing module,setting and retrieval module.Image storage module which can be entered in the U disk by the user and connected to the system by way of mount;Pre-processing module is to extract feature of the query image and database image retrieval,and then save feature into the feature library indexed,it is the set of all image processing algorithms of this thesis,mainly to complete the image retrieval process of image processing.It is a collection of all image processing algorithms,mainly to complete the image retrieval process in image processing.For example,the image color space conversion and quantization,histogram normalized,grayscale image,conversion,GLCM extraction,Gabor transform,gaussian filtering,edge detection and non-maxima suppression;Setting module is to set up some parameters for image retrieval algorithm,to to meet the different needs of users to retrieve;retrieval module is to compare the query image features and image library features by order,calculate the similarity and save the results.HCI module mainly task is to give users easy ways to use graphical interface and control platform by running QT applications,observe and analysis the system's inputs and outputs.With the real-time and accuracy as the evaluation criteria,a full range of tests on image retrieval system have been done.And analysis of the main problems existing in the system has been done,and the solution has been put forward.The test results show that,the system can basically meet the design requirements in both real-time and accuracy.