Design of Machine Vision System for Sugarcane Buds or Rings Detection Akkaranat Rattanaphongphak and Wanwanut Boongsood School of Manufacturing Engineering, Suranaree University of Technology, Nakhon Ratchasima, Thailand Email: a.rattanaphongphak@gmail.com, wanwanut@sut.ac.th Abstract—An important process in sugarcane cultivation is sugarcane preparation and planting using billets with buds and rings. Manually cutting sugarcane into billets causes low quality and productivity for cultivation. Therefore, cutting machines are needed. A proper vision system aids the cutting machine. This study was to design a machine vision program to determine good rings on sugarcane billets. The algorithm was implemented using LabVIEW NI Vision 2015 and a web camera in an enclosed box. The test was performed on sugarcane billets harvested from a farm in Buriram, Thailand. The proposed vision system can determine the sizes of sugarcane billets, colors and locations of the rings. The method for determining rings was evaluating templates using their color spectra in HSL color space. The ring was identified by matching with templates using a color matching algorithm and identified position of rings by color location algorithm. The results of real time testing showed that the machine was able to identify rings correctly at 83.33% and identify the position of rings correctly at 73.81%. Index Terms—agricultural automation, machine vision, sugarcane cutting machine I. INTRODUCTION Sugarcane is the main cash crops in the northeast, the most populous area of Thailand covering 7,601.0736 km2 [1]. Sugarcane propagation is performed using cut rings with root growth. The sugarcane should be cut into shorter segments called billets and they must include a ring. For planting, the rings are the most important part as mentioned above. Nowadays, the billets cutting is done by human labor, or there may be a machine to help; as a result the cutting precision is low, and the speed is low as well, partly because of labor shortage. Therefore, technology in the field of machine vision can be used to identify areas of need for the next processes for sugarcane cultivation. There are many varieties of sugarcane. The specific characteristics of sugarcane must be used in the analysis to design a program to identify areas requiring image analysis. The physical nature of the sugarcane Ref. [2] is typically articulate with different shapes for different species and varieties, but mostly cylindrical. The arrangement of the stalk may be a straight line or a zigzag. The color will vary according to species and the Manuscript received February 16, 2019; revised June 2, 2019. environment. Generally, there are different colors from green to purple to almost black. Different colors are caused by 2 base pigments: green from chlorophyll and red from anthocyanin. Moreover, there is also the bud, which is the most important for the cultivation of sugarcane, as a growth of the roots. The width of this region is not very consistent; the bud side is often wider than the opposite side. From the above, the most important thing to take into consideration to determine the position of ring is the color of the ring area, because it is different from other areas of the stalk, in order to be suitable for use in machine vision to identify this bud region. At present, image processing is being used in the agricultural industry. It must take into consideration both the shape and color of the plants analyzed. The shapes such as protected area, perimeter, ferret diameter and roundness of the plant were identified by the use of color transformation from Red, Green and Blue (RGB) to Hue, Saturation and Intensity (HSI) to separate objects from the background [3]. Moreover, there is also identification of the shapes of the plant species such as apple, strawberry or potato [4]-[6]. The formatter will need to create these components, incorporating the applicable criteria that follow. Image processing has also been used in the sugarcane industry. Sugarcane borer diseases were detected using image segmentation in a grayscale color plane to verify correspondence to the minimum average grey value, filtering and reducing noise to identify borer diseases [7]. The process used three image of the same sugarcane with intervals of 120˚. The method was used to process 50 sugarcane borers images and the accuracy rate was 100%. The billet cut quality has been analyzed for use with a CAMECO harvester for autonomous sugarcane cultivation [8]. The image acquisition was a CCD camera, stereovision camera and a scanning laser rangefinder. The results of image processing to extract a part of the region are the billets, rings and buds of the sugarcane. Before autonomous machines, sugarcane billets were justified and cut by humans. This may cause damage to the rings. This result of the inaccuracy is ring damage, harming the most important area for sugarcane growth. Therefore, the work at present is as follows. In the first section, the image acquisition equipment and methods used to acquire the image of the sugarcane are presented. Finally, program is designed for ring detection and identification of their position. The program for machine Journal of Image and Graphics, Vol. 7, No. 3, September 2019 ©2019 Journal of Image and Graphics 102 doi: 10.18178/joig.7.3.102-106