— Rice is the world’s most important cereal. It constitutes the world’s guideline source of nourishment, being the essential grain for the planet’s most significant populace. It is the staple nourishment and source of dietary vitality and protein. Within the current grain-handling frameworks, grain sort and quality are evaluated by mere visual review. This assessment handle is, be that as it may, dull and time-consuming. Decision-making capabilities can be genuinely influenced by a physical condition such as weakness in vision, the current mental state caused by predispositions and work weight, and working condition such as a disgraceful lighting condition. In this paper, an Android-based image recognition system using Convolutional Neural Network was used as a new method to classify and identify rice variety in terms of visual features such as size, color, shape, and texture of the seeds. Fifty (50) rice samples having eleven (11) rice varieties were used in the study for testing. Results in the study show an overall accuracy rate of 93.8%. Such high accuracy rate confirms that the Android-based Rice Variety Classifier can be used as a tool for classifying rice grains.

selection plays an important role in pattern recognition. The better selection of a feature set usually results the better performance in a classification problem. This work tries to select the best feature set for classification of rice varieties based on image of bulk samples using imperialist competition algorithm. Imperialist competition algorithm is a new evolutionary optimization method that is inspired by imperialist competition. Results showed the feature set selected by the imperialist competition algorithm provide the better classification performance compared to that obtained by genetic algorithm technique

s developed in this study based on image processing techniques to recognize and locate saffron flowers in the field. Color features of the images in HSI and YC rCbcolor spaces were used to detect the flowers. High pass filters were used to eliminate noise from the segmented

food processing and evaluation is considered a significant research area in computer vision. The development of automated cooking and food serving by robots is envisaged as part of automated food processing and temperature plays a major role in cooking Indian foods. The delicious Indian foods are generally boiled or fried with other ingredients. The boiled grains like Bengal Gram, Black Gram, Green Gram, Red Gram and Toor Dal are part of typical Indian foods and taste differently, when boiled or cooked at different temperatures and periods of time. Therefore, identifying the effect of boiling and automatic recognition of images of boiled food grains is presented in this paper. The boiling temperatures chosen are 40 0 C, 50 0 C, 60 0 C, 80 0 C and 100 0 C. A color feature centered knowledge based classifier is proposed. The classification accuracy observed is high at lower and higher temperatures and low at medium temperatures. The work finds applications in automatic inspection of food preparations in food industries, drug preparation in pharmaceutical industries, automatic serving, cooking and monitoring of foods in restaurants and motels.

When distributing rice seed to farmers, suppliers strive to ensure that all seeds delivered belong to the species that was ordered and that the batch is not contaminated by unhealthy seeds or seeds of a different species. A conventional method to inspect the varietal purity of rice seeds is based on manually selecting random samples of rice seed from a batch and evaluating the physical grain properties through a process of human visual inspection. This is a tedious, laborious, time consuming and extremely inefficient task where only a very small subset of the entire batch of the rice seed can be examined. There is, therefore, a need to automate this process to make it repeatable and more efficient while allowing a larger sample of rice seeds from any batch to be analysed. This paper presents an automatic rice seed inspection method which combines hyperspectral imaging and tools from machine learning to automatically detect seeds which are erroneously contained within a batch when they actually belong to a completely different species. Image data from Near-infrared (NIR) and Visible Light (VIS) hyperspectral cameras are acquired for six common rice seed varieties. Two different classifiers are applied to the data: a Support Vector Machine (SVM) and a Random Forest (RF), where each consists of six one-versus-rest binary classifiers. The results show that combining spectral and shape-based features derived from the rice seeds results in an increase in the precision (PPV) of the multi-label classification to 84% compared with 74% when only visual features are used.

Wheat class identification using machine vision is an objective method which can be used for online testing to automate handling, binning and shipping operations in grain industry. The efficiencies of a monochrome camera-based vision system with three different illuminations (incandescent light (IL), fluorescent ring light (FRL), fluorescent tube light (FTL)) were determined to identify eight western Canadian wheat classes at four moisture levels (11%, 14%, 17% and 20%). The monochrome images of the bulk wheat samples were acquired at each moisture level (3 illuminationsx8 classesx4 moisture contentsx100 replications=9600 images). A linear discriminant function was used for the classification of wheat samples using 32 gray level textural features extracted from the monochrome images. The mean gray values of the wheat classes were in the ranges of 75-103, 73-115, and 107-143 for IL, FRL and FTL, respectively. The mean gray values of wheat classes were significantly different within each illumination and between different illuminations (@a=0.05). Mean gray value was the highest for FTL and the lowest for IL illumination. The moisture content of the wheat samples had significant effect on the mean gray values. The overall classification accuracies were 90%, 81% and 96% for IL, FRL and FTL, respectively, when all the wheat classes were at the same moisture levels. It was 66%, 53% and 85% for IL, FRL and FTL, respectively, when the wheat classes were at different moisture levels. The classification accuracies of a 2-stage classification system for the classes with different moisture levels were 68%, 56% and 90% for IL, FRL and FTL, respectively.

Vegetables are widely planted all over China, but they often suffer from the some diseases. A method of major technical and economical importance is introduced in this paper, which explores the feasibility of implementing fast and reliable automatic identification of vegetable diseases and their infection grades from color and morphological features of leaves. Firstly, leaves are plucked from clustered plant and pictures of the leaves are taken with a CCD digital color camera. Secondly, color and morphological characteristics are obtained by standard image processing techniques, for examples, Otsu thresholding method segments the region of interest, image opening following closing algorithm removes noise, Principal Components Analysis reduces the dimension of the original features. Then, a recently proposed boosting algorithm AdaBoost. M2 is applied to RBF networks for diseases classification based on the above features, where the kernel function of RBF networks is Gaussian form with argument taking Euclidean distance of the input vector from a center. Our experiment performs on the database collected by Chinese Academy of Agricultural Sciences, and result shows that Boosting RBF Networks classifies the 230 cucumber leaves into 2 different diseases (downy-mildew and angular-leaf-spot), and identifies the infection grades of each disease according to the infection degrees.

Tomato is one of the important cash crops in the world. There are several varieties of tomato cultivars. Their identifications as well as taxonomy had been studied in the literature using different laboratory methods. Both morphological and/or genetical characteristics were considered in these available studies. However, to the best of our knowledge, there does not exist any study employing an image analysis based approach. Also, morphological features of stomata of tomato cultivars had not been studied before for the present purpose. In this article, we report results of our recent study of morphological features of stomata of different tomato cultivars based on their scanning electron microscopy (SEM) images. Interestingly, these results of the present study are supported by the protein binding pattern of the seeds of respective cultivars.

This research aimed to develop the computer software called “Rice Seed Germination Analysis (RiSGA)” which could predict rice seed image for rice germination by using an image processing technique. The RiSGA consisted of five main process modules: 1) image acquisition, 2) image pre-processing, 3) feature extraction, 4) quality control analysis and 5) quality results. Six variations of Thai rice seed species (CP111, RD41, Chiang Phattalung, Sang Yod Phattalung, Phitsanulok 2 and Chai Nat 1) were used for the experiment. The RiSGA extracted three main features: 1) color, 2) morphological and 3) texture feature. The RiSGA applied four well-known techniques: 1) Euclidean Distance (ED), 2) Rule Based System (RBS), 3) Fuzzy Logic (FL) and 4) Artificial Neural Network (ANN). The RiSGA precision of ED, RBS, FL, and ANN was 87.50%, 100%, 100%, and 100%, respectively. The average access time was 4.35 seconds per image, 5.29 seconds per image, 7.04 seconds per image, and 159.65 seconds per image, respectively.

This paper provides a review of automating or semi-automating the seed quality purity test. Computer vision (CV) technology used in variety of industries is a sophisticated type of inspection technology; however, it is not widely used in agriculture.The application of CV technologies is very challenging in agriculture. As CV plays an important role in this domain, research in this area has been motivated. Several theories of automating seed quality purity test are briefly mentioned. The reviewed approaches are classified according to features and classifiers. The methods for extracting features of a particular seed, and the classifiers used for classifying the seeds, are mentioned in the paper. An overview of the most representative methods for feature extraction and classification of seeds is presented. The major goal of the paper is to provide a comprehensive reference source for the researchers involved in automation of seed classification, regardless of particular feature or classifier.

This paper proposes a solution to localization and classification of rice grains in an image. All existing related works rely on conventional based machine learning approaches. However, those techniques do not do well for the problem designed in this paper, due to the high similarities between different types of rice grains. The deep learning based solution is developed in the proposed solution. It contains pre-processing steps of data annotation using the watershed algorithm, auto-alignment using the major axis orientation, and image enhancement using the contrast-limited adaptive histogram equalization (CLAHE) technique. Then, the mask region-based convolutional neural networks (R-CNN) is trained to localize and classify rice grains in an input image. The performance is enhanced by using the transfer learning and the dropout regularization for overfitting prevention. The proposed method is validated using many scenarios of experiments, reported in the forms of mean average precision (mAP) and a confusion matrix. It achieves above 80% mAP for main scenarios in the experiments. It is also shown to perform outstanding, when compared to human experts.

This paper presents a system for automated classification of rice variety for rice seed production using computer vision and image processing techniques. Rice seeds of different varieties are visually very similar in color, shape and texture that make the classification of rice seed varieties at high accuracy challenging. We investigated various feature extraction techniques for efficient rice seed image representation. We analyzed the performance of powerful classifiers on the extracted features for finding the robust one. Images of six different rice seed varieties in northern Vietnam were acquired and analyzed. Our experiments have demonstrated that the average accuracy of our classification system can reach 90.54% using Random Forest method with a simple feature extraction technique. This result can be used for developing a computer-aided machine vision system for automated assessment of rice seeds purity.

This paper presents a neural network approach for classification of rice varieties. A total of 9 different rice verities were considered for the study. Samples were drawn from each variety and images of seeds were captured. Algorithms were developed to extract thirteen morphological features, six colour features and fifteen texture features from colour images of individual seed samples. A different neural network models were developed for individual feature sets and for the combined feature set. High classification accuracy was given by textural features than morphological and colour features. An overall classification accuracy of 92% was obtained from combined feature model. Individual classification accuracies of AT307, BG250, BG358, BG450, BW262, BW267, W361, BW363 and BW364 were 94%, 98%, 84%, 100%, 94%, 68%, 98%, 94% and 94% respectively. It was noted that different neural network architectures tend to produce different accuracies.

This paper presents a method for classification of vegetables based on RGB colour and local binary pattern (LBP) texture features. The feature vector comprises of the combination of colour and texture features that contribute to the classification. Leafy and non-leafy vegetable images are deployed. In this work 18 varieties of vegetables are considered by choosing nine leafy and nine non-leafy vegetables. A multilayer neural network is used for the classification. The experimental results demonstrated that, with neural networks classifier an overall classification accuracy of 93.3% is achieved across different vegetables. The work finds useful in developing recognition system for super market, packing and grading of vegetable, food processing and Agriculture Produce Market Committee (APMC).

This paper presents a machine vision application designed for rice seed germination analysis by using image processing and computer vision technology. The application is called “Rice Seed Germination Analysis (RSGA)". RSGA consists of five main processing modules which are image acquisition, image preprocessing, feature extraction, quality control analysis and quality results. The experiments is conducted on six variation Thai rice seed species of CP111, RD41, Chiang Phatthalung, Sang Yod Phattalung, Phitsanulok 2 and Chai Nat 1 in Bangkok and Chiangmai province of Thailand. RSGA extracts four main features which are color, size, shape, and texture. Then, RSGA applies Artificial Neural Network techniques in crop germination prediction. The precision rate is 93.06 percent, with the speed 8.31 seconds per image. Introduction Thailand is one of agricultural countries which produces a hugh number of food products each year. Thailand is one of agricultural countries which produces a huge number of food products each year. Nowadays, the agricultural industry is probably more widespread in the world. One product that is highly widespread in the world especially in Thailand is rice. Oryza Sativa (Rice) is a vital worldwide agricultural product. Rice is a very popular exported product in the world. In Thailand, there are more than 114 well-known Thai rice species. It is essential to grade the quality of commodities in order to command the better price in the market competitions. The quality of the rice is mostly based on the quality of the seeds. Paddy rice is mostly offered based on the good quality of products. To measure the seed quality assessment, many factors is considered in the germination test, for instance, seed quality additives, species purity, physical purity, insect pests and diseases, seed germination and seed vigor. However, it is very difficult to identify which kind of seeds should be better used in seed germinations. Thus, this research is applied the image processing and computer vision technique instead of using only human vision. The traditional way to conduct the germination test is based on human abilities, it takes times and high labor to conduct the germination test in the seed quality control process. Nevertheless, it is very difficult to identify the quality of rice seeds by using only human vision because it is time-consuming and uses high labor to assess the quality control process in order to earn benefits from the rice productive. The germination test in this research was applied following the standard seed germination test in ISTA (1996) with the top of paper method for crop germination. Moreover, images have been collected to predict the germination of rice seed images by using image processing techniques which can identify the quality of products. Therefore, the objective of this research is to propose the novel technique by applying the image processing and computer vision technology to assess the crop germination prediction in order to reduce times and costs. Thus, the computer software which can predict rice seed image for crop germination by using image processing techniques is developed. Due to the advance of video camera technology, people can take a digital picture or digital video stream easily in any places and any time by a camera or by a mobile phone device. It is not only very easy to use a digital camera, but also it is inexpensive. Moreover, it is easy to transform and process by using a computer system. Thus, this research employs a digital camera to capture the image. Many researches have been conducted the rice seed classification and rice seed recognition systems based on varieties, shape and size [7-8, 11, 13-16, 18, 20-22, 24]. The system of this research aimed to allow users to load an unknown whole rice seed image into RSGA which the system attempts to predict which kind of seeds were germinated or non-germinated for crop germination. Finally, the RSGA displays the germination results on the system’s graphic user interface (GUI). Many experiments were conducted to evaluate the seed germination by using machine vision technology. Related Work Many researchers tried to classify and identify each rice seed by applying several techniques which can be classified into two categories: 2.1) Computer techniques and 2.2) Biological techniques. These below details show the information relating to the techniques used in this research. For computer techniques, Pornpanomchai [17] classified pattern recognition approaches into three main categories: 1) Statistical pattern recognition (StatPR), 2) Structural or syntactic pattern recognition (SyntPR) and 3) Neural-based pattern recognition (NeurPR). The details of each category are as follows: International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 9 (2016) pp 6800-6807 © Research India Publications. http://www.ripublication.com 6801 Statistical pattern recognition (StatPR) This approach has a statistical basis for classification of algorithms. The classifier design attempted to use statistical and a priori probabilities to build the recognition algorithms. The input features were extracted from a set of pattern characteristic measurements. Patil and Yadahalli [16] applied minimum distance and knearest neighbor for classifying 21 classes of different food grains by using color and texture features without performing preprocessing and segmentation. The precision rate was 83.66%. Pandey et al. [15] used content based image retrieval (CBIR) technique for identifying different seeds (wheat, rice and gram). The CBIR technique was used to identify and recognize the images based on color, shape, texture and size features. Finally, Euclidean Distance (ED) and artificial neural network techniques were applied in order to do seed identification. The training data set contained 150 images and the testing data set contained 50 images. The precision rates were 84.4% for ED and 95% for artificial neural network. Kaur and Singh [10] applied Multi-Class SVM to grade the rice kernels (Premium, Grade A, Grade B and Grade C) for different varieties of rice grains based on interior and exterior quality. Maximum variance method was used to extract the rice kernels from background and extract chalk from rice. Ten geometric features were used to determine the percentage of head rice, broken rice and brewers of the rice samples. The precision rate was more than 86%. Structural or syntactic pattern recognition (SyntPR) This approach was a structural entity used for classification and descriptions. The classification was based on measurements of structural pattern similarity. The structural or syntactic information was used to generate knowledge related to patterns by extracting the similarity of patterns in order to build pattern syntax or structural rules. The information of pattern syntax rules were used to explain, classify and recognize unknown patterns. This approach was classified into two subcategories which were Rule Based Expert Systems and Fuzzy Expert Systems. Punthimast et al. [34] applied a rule based system to identify rice seeds and stick rice seeds based on RGB color histogram. The precision rate was 96.34% for “Jasmine” rice seeds and 100% for stick rice seeds. Maheshwari and Jain [26] proposed a method for counting the number of Oryza sativa L (rice seeds) with long seeds as well as small seeds by comparing the degree of quality and quantifying these degrees for the rice seeds based on combined measurements. They applied a rule based system to construct the rules based on histogram of area, major axis, minor axis and eccentricity of the seeds in order to classify the rice seeds into normal, long and small seeds. Finally, they compared the results with the ground truth table based on Human Sensory Panel of various sample. Ajay et al. [1] proposed an algorithm of evaluating the automatic evaluation method for the determination of the quality of milled rice seed based on shape and geometric features. They applied rule based system of the length of the rice kernel to classify the broken rice and non-broken rice by using morphological features applied with the minimum bounding rectangle method. It was proved that this method was efficient. Moreover, this method was able to improve the traditional one. Sansomboonsuk and Afzulpurkar [20] developed a computer system for evaluating the quality of rice kernels. The system applied fuzzy logic to organize and classify the class of each kernel into broken rice and long grain rice by using the point and line touching features (area, perimeter, circularity and shape compactness). A rule based system was applied to generate the fuzzy set of rules in order to classify each rice kernel. The precision rate was 90% for evaluating the quality of rice compared with human inspection methods. Neural-based pattern recognition (NeurPR) This approach was a nonalgorithmic or black box strategy which was trainable. This approach also fed input features into neural network nodes to identify patterns. The NeurPR emulated knowledge of how biological neural system stored and manipulated information. The artificial neural system called “neural networks” could solve the problems of automatic reasoning including the pattern recognition problem. This approach classified patterns by predicting the properties of neural network. Guzman and Peralta [8] applied artificial neural networks to classify the grain samples of Philippine rice grains based on size, shape and 52 varieties of rice grains belonging to 5 varietal groups in the Philippines by using image processing techniques. The system used 3 data sets containing 110 rice grains for each size, shape and variety of rice grain identification. They achieved a precision rate of 98.76% for sizes and 96.67% for shape. The precision rate based on individual rice varietal types were 85.81, 94.58, 96.16 and 97.39% for the lowland irrigated, lowland

This paper investigated the use of hyperspectral imaging (HSI) to discriminate the variety and quality of rice. Hyperspectral images (400–1,000 nm) of paddy rice samples were acquired to extract both spectral and image information. Dimension reduction was carried out on the region of interest (ROI) of the images by principal component analysis (PCA). The first principal components (PCs) explained over 98 % of variances of all spectral bands. Chalkiness degree and shape feature (‘MajorAxisLength’, ‘MinorAxisLength’, length-width ratio, ‘Perimeter’ and ‘Eccentricity’) were further extracted and used for subsequent rice variety discrimination by PCA and back propagation neural network (BPNN). An integration of spectral and image data was used for BPNN classification. The BPNN model based on spectral data (seven optimal wavelengths) achieved better results than PCA based on spectral data (seven optimal wavelengths) in variety discrimination with classification accuracy of 89.18 and 89.91 % for PCA and BPNN model, respectively. The BPNN model based on data fusion achieved the best results (94.45 %), which was superior to the results based on spectral data (89.91 %) or image data (88.09 %) alone. Finally, the resulting classification maps were able to visualize different rice varieties. The results demonstrated that discrimination of rice variety and quality with HSI technology was feasible and could be utilized for quality control purposes and/or for innovative sorting of rice.

This paper describes the classification of four different varieties of rice grain based on four sets of features, namely morphology, colour, texture and wavelet. The classification is carried out on single rice kernel using image pre-processing steps followed by a cascade network classifier. The performance of the classifiers based on the above feature sets is also compared. It is found that morphological feature is more suitable for the classification of rice kernels, as compared to other features. The number of input features is reduced by a feature selection process using statistical analysis system (SAS) software. The classification accuracy based on selected features is compared with that of original features using different classifiers. It is found that the selected features are able to provide classification accuracy very close to the original features. The performance of the proposed cascade classifier is also tested against standard datasets from the University of California, Irvine (UCI), and the results are compared with other classifiers. The results show that the proposed classifier provides better classification accuracy as compared to other classifiers.

This paper describes a new simple logical method for finding perspective rectangle structural shapes on the image. The proposed method uses a one (1) dimensional array to analyze the rectangular shape. Today machine vision is a useful tool for external feature measurement since it provides a reliable and objective means for evaluating products based on visual features. An ultimate goal of this project is to trap the rectangular shape objects. This paper focuses on one main primitive operation, i.e. shape analysis.

This paper deals with the classification of eight types of rice grain using image processing and neural network. Three different texture feature extraction schemes based on co-occurrence matrix, run length matrix and wavelet decomposition were considered. The contribution of these texture feature extraction techniques towards rice grain classification were analysed and compared. A back propagation neural network is used for this classification task. The performance in terms of classification accuracy of the above three texture feature extraction schemes where tested. It is found that texture feature based on wavelet decomposition is able to classify eight different types of rice grain with an overall classification accuracy of 98.87% as compared to other texture feature extraction schemes discussed in this paper.

This paper aims to present a review of seed technology, seed germination and vigor methods using image processing. Computer–aided image analysis techniques have been recently developed in monitoring seed growth and vigor. Their integration with the standard germination test is needed to describe the germination performance of a seed sample with high accuracy. The use of various modern image acquisition techniques combined with image processing techniques have allowed developing automated seed quality tests,. The two main limitations of performing a vigor test manually are 1) results of a vigor test may vary from laboratory to laboratory because of the subjective nature of most vigor tests and 2) many vigor tests take excessive time to acquire results. These two limitations can be addressed by designing computer software that measures the seedlings represented by a digital image and computes the vigor index from those measurements. Several theories of seed germination and vigor are briefly mentioned. Methods are classified into several groups. Keywords—Artificial Vision System, Living Ability Index, Seed Germination, Seed Vigor, Seed Quality Tests, Vigor