Study On Image Processing Material Classification System Based Embedded Arm

Material is to be divided into different grades according to their characteristics and this process is named material classification. The quality grading of agricultural products is mainly dependent on their exterior features. Classification can be divided into artificial classification and mechanical classification. Artificial classification is Inefficient and the classification standards are not uniform. And early mechanical grading is effective when hard material is used but fails to deal with agricultural products of which the shell is easy to be damaged. With the development of optical image processing technology in recent years, grading methods based on new machine vision have great prosperity for the colorful, defective and vulnerable materials. Since China's accession to WTO, agriculture is facing competition in the world. Agricultural products which have been strictly graded are beneficial to be sold in the international market.The paper has proposed a material classification system based on machine vision which is mainly made up of ARM and a camera for image processing with the purpose of static and automatic grading of agricultural products according to their size. The hardware is mainly composed of transportation equipment, camera systems, embedded processors and grading device and the software is to pick image and identify the material size. When images are picked and stored in the ARM system, the corresponding material characteristics are extracted from the images following some necessary pre-processing and then grading level is determined.The image acquisition and recognition system introduced in this paper consists of embedded ARM (LPC2103) and C328 camera pose. Under the control of the ARM system, the mode of 8bit gray image of C328 is selected to obtain hexadecimal dates which only include color component and is not compressed according to the requirement of specific identification object with the purpose of occupying the memory as little as possible and speeding up the recognition speed under the condition of complete of recognition. And then image is undergone binary processing and purification. Furthermore, the following three kinds of algorithm are used to calculate the image size: (1) calculation of cross-sectional area of material image. (2) calculation of the average radius by obtaining the center of the circle (3) obtaining the largest diameter by scanning method. As a result of comparing and analyzing the repeated experimental results, the algorithm (2) which has a recognition error less than 5% and superior recognition speed is considered to be better than other two algorithms. The paper also briefly describes the composition and principle of the coupled step-forward grading system controlled by image recognition system. The combination of mechanical and electrical integration makes up a practical classified machine of potato.The system using relatively economic ARM and camera that we introduce in this paper has made up of the core of the practical grading system which can theoretically classify material according to size randomly and makes a good foundation for the agricultural products grading by color or defect.