Improved Design And Experimental Study On Hami Melon Classifier Based On Line Scan Digital Camera

The external quality of Hami melon include size, shape, color, texture, and other defects. Comprehensive classification of external qualities for Hami melon based on the machine vision, not only can reduce labor costs and improve the efficiency of classification, also can achieve higher prices for better quality of Hami melon commodity, and improve the market competitiveness of Hami melon. In this paper, for the original Hami melon classifier, t three main research works were conducted, including the improvement of hardware, the optimiza t io n of software, the optimization of software system, and the module establishment and implementation of classification algorithm of Hami melon, based on the size, color, and texture features.The main research contents and conclusions are as follows,(1) Hardware Improvement of Hami melon classifier. Acquisition device, the output device were separated from the main body of the classifier, for reducing effect of vibration on image acquisition, and improving image quality, image segmentation and feature extraction.(2) Optimization of grading software of Hami melon. It is rewrited about the PC program and man-machine interface, including new login module, data storage module, image processing and analysis module.(3) The establishment of grading module of Hami melon. The extraction method is proposed for external quality characteristics of Hami melon, such as geometric size, color, texture. The color, size and surface texture grading modules of Hami melon were put forward sing the color area percentage of body size in terms of the H component image, region el ipse fitting, and gray level co-occurrence matrix respectively. And the accuracy of grading with three characteristics were respectively 89.14%, 91.34% and 93.75%. To integrated three features to establish the evaluation model of Hami melon, the accuracy rate of classifica t io n was up to 84%.(4) Performance test of Hami melon classifier. In the performance test using validat io n sample, classification correct rate of Hami melon classifier was 86.67%.