The Realization And Research On The Method Of Mature Orange Recognition And Trunk Contour Extraction

With the rapid development of agricultural production, the rapid rise of theagricultural labor costs and the acceleration of the trend of population aging, thelack of agricultural labor force will become the common problems faced by manycountries.Mechanization and automation of the orchard harvest operations becomeone of the most hot issues. Researching on the key technology of fruit picking robotcan not only meet the demend of the market and reduce labor intensity,but alsokeep up with the pace of the new developing agrieultural technology in the modernworld and improve economic effieieney. At the same time，it is important forimproving Chinese agrieultural modernizationHow to recognize the target accurately and in real time is a key issue forIntelligent fruit picking robot. In this study which is a part of research on matureOrange harvesting machine vision system, A method based on color features isproposed based on the analysis of the relevant research present situation of thedomestic and foreign to identify mature Orange and Orange trunk in the naturalscene. The main contents and methods of this study are as follows:(1)A method of adaptive binarization to extract the shape and trunk contour ofOrange trees is used. firstly, a picture of Oranges is entered and the original imageis transformed into a grayscale image, then the picture is processed by adaptivebinarization and Median filtering to remove the background noise in the image, thepicture is processed by morphological processing, setting a threshold value for thetarget contour is to screen the contours in Binary image. The contours can beextracted well through the method and provides information about the target forpicking robots.(2)A method based on color feature for mature Oranges is proposed. Usingcolor fetures to identify mature Oranges. The basic principle of the method is toextract the histogram of the color feature of the fruit samples. the color with similarcharacteristics is found and matched in order to identify mature Oranges bymarking characteristic color of fruit and counting characteristics of the colorhistogram.Because of the characteristics of mature Oranges and the colordifferences between background and Oranges, The method can remove thebackground noise in the image effectively and the rate of mature Oranges recognition is high.The experiment show that the method can identify mature Oranges moreaccurately and achieved the objective of the study through Collection of naturalscene pictures of experiment and simulation and the expected demand, and providea basis for picking robot.