Detection of green apples in natural scenes based on saliency theory and Gaussian curve fitting

Abstract

Green apple targets are difficult to identify for having similar color with backgrounds such as leaves. The primary goal of this study was to detect green apples in natural scenes by applying saliency detection and Gaussian curve fitting algorithm. Firstly, the image was represented as a close-loop graph with superpixels as nodes. These nodes were ranked based on the similarity to background and foreground queries to generate the final saliency map. Secondly, Gaussian curve fitting was carried out to fit the V-component in YUV color space in salient areas, and a threshold was selected to binarize the image. To verify the validity of the proposed algorithm, 55 images were selected and compared with the common used image segmentation algorithms such as k-means clustering algorithm and FCM (Fuzzy C-means clustering algorithm). Four parameters including recognition ratio, FPR (false positive rate), FNR (false negative rate) and FDR (false detection rate) were used to evaluate the results, which were 91.84%, 1.36%, 8.16% and 4.22%, respectively. The results indicated that it was effective and feasible to apply this method to the detection of green apples in nature scenes. Citation: Li B R, Long Y, Song H B. Detection of green apples in natural scenes based on saliency theory and Gaussian curve fitting. Int J Agric & Biol Eng, 2018; 11(1): 192-198. Li B R, et al. Detection of green apples in natural scenes based on saliency theory and Gaussian curve fitting Vol.

A comprehensive review of fruit and vegetable classification techniques

“…The study reported that the experiments were not performed on a uniform dataset and the result are not comparable with the state-of-the-art. To detect the green apple a graph based manifold saliency was used with k-mean and Fuzzy-C-Mean (FCM) clustering, where the study reported on imperfect segmentation that needs to be integrated via an area loss function [14]. A more related research has been presented for quality evaluation of packed lettuce, where a patch based segmentation has been performed with CNN.…”

“…A non-exhaustive description of shape, texture and colour feature descriptors has been described in this section. 1996 Mixed Classification Threshold-based pixel level image subtraction [66] 2006 Apple Quality assessment Feature-based with variable neighbourhood size [30] 2006 Citrus Quality assessment Texture based HSI and Colour Co-occurrence (CCM) [31] 2007 Apple Quality assessment Gabor kernel and PCS avoided local features segmentation [32,108] 2012 Mixed fruit Fruit harvesting Spatial-local adaptive threshold based [52] 2012 Mixed Classification Distance Transform (DT) and watershed [77] 2013 Vege Detection Texture and edge fusion segmentation [78] 2015 Mixed fruit Detection K-mean split and graph-based merge with area threshold [110] 2016 Apple Recognition Dynamic threshold Otsu method [111] 2016 Mixed fruit Classification Square window split and merge segmentation [72] 2016 Tomato Quality assessment Otsu method [39] 2017 Apple Bruise detection HSI based Otsu method [40] 2017 Eggplant Grading Intensity adaptive threshold based Otsu [57] 2018 Apple Detection Graph based k-mean FCM clustering [14] 2018 Litchi Robotic harvesting One dimensional random signal histogram with FCM [59] 2018 Mixed fruit Detection Fusion of LBP, HOG, global colour and shape with Otsu [7] 2018 Packed food Quality assessment 3 × 3 patch likelihood threshold with CNN [109] 2018 Papaya Disease detection K-mean clustering based segmentation [44] 2018 Pomegranate Clustering Threshold Otsu [81] Ability to recognise and retrieve from partial information.…”

“…A summary of recent fruit and vegetable classification performed in different real-life applications is presented in Table 1. Recent state-of-the-art for fruit and vegetable classification and recognition are a combination of feature description and machine learning algorithms on visual data [1,12,13,14,15]. Significant research has been reported for representation of different characteristics of fruit and vegetable as feature vectors [6,16].…”

A Sample Weight and AdaBoost CNN-Based Coarse to Fine Classification of Fruit and Vegetables at a Supermarket Self-Checkout

“…The vision-based classification of fruit and vegetables has been performed in many fields for a range of different applications. The most common applications include the classification of fruit or vegetables for automated harvesting in agricultural settings [18][19][20] or vision-based quality assessment of fruit or vegetables [21][22][23].…”

Preprocessing method of night vision image application in apple harvesting robot

“…In general, there are two factors for improving harvesting efficiency under the unstructured environment of an orchard. One is the self-performance improvement of apple harvesting robots, such as the monocular vision [11] and binocular vision [12] , recognition and tracking of dynamic fruit [13] , manipulator obstacle avoidance [14] , path planning [15] , green apple recognintion [16] , and so on. Another one is longer working time to implement the round-the-clock operation.…”