An automatic segmentation method for multi-tomatoes image under complicated natural background

Abstract

It is a fundamental work to realize intelligent fruit-picking that mature fruits are distinguished from complicated backgrounds and determined their three-dimensional location. Various methods for fruit identification can be found from the literatures. However, surprisingly little attention has been paid to image segmentation of multi-fruits which growth states are separated, connected, overlapped and partially covered by branches and leaves of plant under the natural illumination condition. In this paper we present an automatic segmentation method that comprises of three main steps. Firstly, Red and Green component image are extracted from RGB color image, and Green component subtracted from Red component gives RG of chromatic aberration gray-level image. Gray-level value between objects and background has obviously difference in RG image. By the feature, Ostu's threshold method is applied to do adaptive RG image segmentation. And then, marker-controlled watershed segmentation based on morphological grayscale reconstruction is applied into Red component image to search boundary of connected or overlapped tomatoes. Finally, intersection operation is done by operation results of above two steps to get binary image of final segmentation. The tests show that the automatic segmentation method has satisfactory effect upon multi-tomatoes image of various growth states under the natural illumination condition. Meanwhile, it has very robust for different maturity of multi-tomatoes image.

Fuzzy classification of pre-harvest tomatoes for ripeness estimation – An approach based on automatic rule learning using decision tree

“…These techniques cannot be utilized for estimating the ripeness of tomato before harvesting. Some researchers proposed the algorithm for tomato recognition that has been done under natural background also but it did not take into the account the expected highlights on the tomato surface due to natural illumination conditions [20,21]. The white pixels of specular highlights on the tomato surface gets removed during segmentation process which results in loss of color information.…”

“…Authors Yin et al [21], proposed an automatic segmentation technique based on RG difference of an image for segmenting multi-tomatoes under complicated natural background. Authors Xiang et al [20], examined and relatively evaluated three segmentation algorithms based on RG, normalized RG, and multi-band ratio.…”

Elaeis Guineensis leaf image segmentation: A comparative study and analysis

“…Here, the image space with similar intensity values were partitioned into clusters of regions. Next, Yin Jianjun et.al [8] proposed the automatic segmentation method of segmenting multi-tomatoes image of various growth states. The marker-controlled watershed segmentation based on morphological grayscale reconstruction was implemented to search boundary of connected or overlapped tomatoes.…”

A Vision System Based on TOF 3d Imaging Technology Applied to Robotic Citrus Harvesting

“…Color properties and shape properties of color image were extracted, and the different classification methods were used for recognition of fruit (Huang, He, & Yang, 2013;Lak, Minaei, Amiriparian, & Beheshti, 2010). The methods based on circular Hough transform (Xie, Zhang, & Zhao, 2007;Cai, Zhou, Li, & Fan, 2008;Yao, Ding, Zhao, & Yang, 2008), watershed algorithm (Yin, Mao, Wang, Chen, & Zhang, 2006;Zhang, Lin, & Gao, 2004;Zhou, Zhang, Yang, & Zhao, 2007), least-squares circle fit (Lü, Cai, Zhao, Wang, & Tang, 2010) and geometry analysis (Plebe & Grasso, 2001;Gu, Lu, Lou, & Zhang, 2006;Cai, Wang, Chen, Wang, & Lü, 2009) were proposed to solve the problem of shading and overlap. It can perform well when a good contour image is obtained, but false detections also occurred due to circular contours generated by edges of leaves and branch.…”