Analysis of Sorting Fresh Jew's Mallow (Corchorus Olitorius) Leaves Using Imagery Characteristics

摘要

Development in the computer world has led to the solution of many problems in the agricultural field. Classification and sorting of vegetables are the complex fields that require great human experience. Emergence of new technologies can contribute forproblem solving, such as machine vision. As Jew’s mallow is a leafy vegetable that is very sensitive to environmental conditions and rapidly deteriorate after harvesting, so it is necessary to configure out a suitable proof of concept for that visual sorting in processing plants. In the present study, several Jew’s mallow’s leaves were selected, varying widely in terms of severity of greenness, which varied due to different circumstances. An optical meter was used to measure the Chlorophyll ContentIndex, CCI, for each leaf. Experiments were carried out in November 2019 at the Laboratory of Department of Horticulture, Faculty of Agriculture,‘Kafrelsheikh University. The experimental procedures were performed in three stages. In the first stage, aCCI-based primary classification is made that is further converted to wilting percent. In the second stage, a digital Red, Green, and Blue (RGB) camera is used for image capturing for each leaf sample class. Then a scale for leaves images from freshnessto wilting was established according to General Appearance (GA) rules and wilting percent. After which the classified leaves images of Corchorus olitorius are processed to extract characteristic features among the predetermined classes. Morphological analysis shows a significant difference among classes and RGB pixel intensity distribution on grey scale from 0 (pure white) to 255 (pure black) can potentially differentiate among classes according to RGB intensities on the scale. Based on the statisticalanalysis of pixel intensity distribution of each image, the developed non-linear multiple regression equation (R2=0.99) could predict precisely the wilting percent of the leaf. Eventually the color gradient model can be used effectively to discriminate the leaf color of green, yellow and dark.