The Rice Quality Detection Based On Machine Vision Research And The Dsp Implementation

The measurement of exterior quality of rice in our country is still based on manualinspection nowadays, which is inefficient and lack of consistency. In response to theseshortcomings, this thesis developed a Rice-Quality-Detection system based on MachineVision. In this system, ARM board is used as the control kernel which in charge ofsystem control, communication between the boards, etc. The Texas Instruments (TI)DSP--DM642works as main processor for computing and operating the correspondingalgorithms. This system can receive real-time analog video signal from video port, andthrough processing can achieve the purpose of quality inspection of rice. The relevantindicators and data can be transmitted to PC&ARM board through the Serial port. Also,the user of the system can change parameters and view the real image processing resultdirectly.This thesis’s mainly works include two parts, which are the study&choice ofalgorithm and the DSP complement&optimization.(1) The study and choice of algorithm. Simulation of most commonly used digitalimage processing algorithms, including filter de-noising, edge detection, automaticthreshold segmentation etc, through comparative analysis, chose appropriate basicimage processing algorithm that fit our application; Using color space transformationand chromaticity histogram tools, studied the yellow rice detection method based onchrominance information. The simulation results shows that this method can distinguishyellow rice correctly; Considering the limitations of rectangular coordinates, study onpolar coordinates has been carried out. Using this method, we can easily get severalimportant methods of rice measurement: centroid detection, edge detection and long-short-axis detection etc. according to the standard we predetermined, simulation resultshows that different of rice types or broken rice can be found by these methods.(2) The DSP complement&optimization. Finished the whole software framedesign and write the function code with C language; Using series of board-leveloptimization methods, like Chip Support Library (CSL) DAT module, Video driver etc,accelerated the data transmission rate so that the whole system speed s up10times; Algorithm Optimization and CCS’s C-complier is used, which make the compilationand optimization of C-codes seamless and more efficient; C-code optimizing, such asdata type choice, loop unrolling, data table operation and so on, carried out so thatparameters of the system are further improved10times and achieved the projectrequirement.