Classifying Textile Designs Using Bags of Shapes

Shape descriptors have demonstrated encouraging potential in retrieving images based on image content. A number of shape descriptors have been reported in the literature. Nevertheless, most of the reported descriptors still face accuracy and computational challenges. Fourier descriptors are considered to be promising descriptors as they are based on sound theoretical foundation, and possess computational efficiency and attractive invariance properties. In this paper, we propose a novel Fourier descriptor based on contour curvature. The proposed descriptor takes an unconventional view of the curvature-scale-space representation of a shape contour as it treats it as a 2-D binary image (hence referred to as Curvature-Scale Image, or CSI). The invariant descriptor is derived from the 2-D Fourier transform of the Curvature-Scale Image. This allows the descriptor to capture detailed dynamics of the shape curvature and enhance the efficiency of the shape matching process. Experiments using images from the MPEG-7 database have been conducted to compare the performance of the proposed descriptor with the Curvature- Scale-Space Descriptor (CSSD), the Generic Fourier Descriptor (GFD), and the 1-D Fourier Descriptor (1-FD). The proposed descriptor demonstrated superior performance.

There are various possible approaches to the task of automatic identification of football players on video sequences. In the paper the numbers on jerseys are analysed and investigated. Firstly, the algorithm for the localisation of sport shirts and consecutively numbers is described. Secondly, the method for the recognition of the extracted numbers is proposed and experimentally investigated. The approach analysed in the paper applies the Generic Fourier Descriptor for the representation of extracted binary shapes of numbers.

Content Based Image Retrieval (CBIR) is important in computer aided plant species recognition. Texture and shape information have been the primitive image descriptors in content based image retrieval systems. This paper presents a novel framework for combining both the features, texture and shape information, and achieve higher retrieval efficiency. The study provides a methodology for retrieving medicinal plants images from a database of medicinal plant images based on shape and texture features. The shape descriptors include Zernike moment, Fourier descriptor (FD), Generic fourier Descriptor(GFD) and for texture descriptors gabor filters are used. The similarity measures, euclidean distance of each medicinal plant image from the database to query image is used. The images are sorted based on similarity of Euclidean distance. The retrieval experiments are carried on different training and test medicinal plant images. The effectiveness of different descriptors is confirmed by the experimental results. We have investigated shape and texture features for medicinal plant retrieval by successively combining the different transforms. The retrieval efficiency is reported through precision and recall rate. Experimental results by combining Gabor and Zernike transform outperforms the all other methods.

In this paper, we propose a method for indexing and retrieval of images based on shapes of objects. The concept of connectivity is introduced. 3D models are used to represent 2D images. 2D images are decomposed a priori using connectivity which is followed by 3D model construction. 3D model descriptors are obtained for 3D models and used to represent the underlying 2D shapes. We have used spherical harmonics descriptors as the 3D model descriptors. Difference between two images is computed as the Euclidean distance between their descriptors. Experiments are performed to test the effectiveness of spherical harmonics for retrieval of 2D images. The proposed method is compared with methods based on principal components analysis (PCA) and generic Fourier descriptors (GFD). It is found that the proposed method is effective. Item S8 within the MPEG-7 still images content set is used for performing experiments.

Sign languages are used all over the world as a primary means of communication by deaf people. Sign language translation is a promising application for vision-based gesture recognition methods. Therefore, it is need such a tool that can translate sign language directly. This paper aims to create a system that can translate static sign language into textual form automatically based on computer vision. The method contains three phases, i.e. segmentation, feature extraction, and recognition. We used Generic Fourier Descriptor (GFD) as feature extraction method and K-Nearest Neighbour (KNN) as classification approach to recognize the signs. The system was applied to recognize each 120 stored images in database and 120 images which is captured real time by webcam. We also translated 5 words in video sequences. The experiment revealed that the system can recognized the signs with about 86 % accuracy for stored images in database and 69 % for testing data which is captured real time by webcam.

Face detection is an important part of any face recognition system and it is a difficult task to detect faces of arbitrary size, orientation and location. Fusion is a technique which combines the results from two or more distinct methods in such a way that better results are obtained. We propose two different fusion techniques, data fusion and sensor fusion, that combine statistical and shape based methods. Four different features, eigenweights from Principal Component Analysis (PCA), Legendre Moments (LM), Zernike Moments (ZM) and Generic Fourier Descriptor (GFD), are used for our fusion methods. In data fusion technique, the calculated eigenweights, moments, and descriptors are concatenated together to form a feature vector. This combined feature is then classified as face or non-face pattern using Support Vector Machine (SVM). In the case of sensor fusion technique, each of the feature vectors obtained by individual methods are first classified using individual SVMs. Then the class label obtained from the individual SVMs are concatenated together to form a feature vector. This feature vector is then classified using another SVM. Both the techniques produced better results compared with the individual methods. For a database containing 300 face images and 200 non-face images, both of our methods achieved 0% false rejection rate. 99.75% accuracy was achieved for data fusion method, while sensor fusion method produced 99.9% accuracy.

We propose a method for enhancing the accuracy of shape descriptors. The concept of connectivity to obtain discriminating shape descriptors, is introduced. We show how connectivity is applied to two popular shape descriptors. Experiments are performed to test the effect of using connectivity with generic Fourier descriptors and distance histograms. Item S8 within the MPEG-7 still images content set is used for performing experiments. This dataset consists of 3621 still images. The experimental results show that connectivity enhances the performance of the methods significantly.

Textile image segmentation is widely used in textile industry design, since users often need to reconstruct and redesign the patterns of the textile image. Different from traditional image segmentation methods, this paper focused on handling textile images, which received little attention until now. Taking into account the characteristics of textile, this paper proposed a novel graph theory and region merging strategy based textile image segmentation method. Our method first generated the over-segmented image by applying the graph-based image segmentation on the original image. Then we extracted the predominant color to mark the background segments. The region action graph was proposed to improve the conventional region adjacency graph before building the region relation graph for the following region merging. It can greatly improve the segmentation quality since textile image usually includes the regions with complex distribution of different colors. In the phase of region merging, we formulated it as designing merging criterions for the relate regions with geometry properties, such as globalist, locality, and spatial continuity. Extensive experiments were performed and the results showed that our method can reliably segment the textile images into sections with perceptual meaning. Additionally, our method is simple and efficient, with great potential in practical applications.

In the context of microbiology, recent studies show the importance of ribonucleo-protein aggregates (RNPs) for the understanding of mechanisms involved in cell responses to specific environmental conditions. The assembly and disassembly of aggregates is a dynamic process, the characterization of the stage of their evolution can be performed by the evaluation of their number. The aim of this study is to propose a method to automatically determine the count of RNPs. We show that the determination of a precise count is an issue by itself and hence, we propose three textural approaches: a classical point of view using Haralick features, a frequency point of view with generalized Fourier descriptors, and a structural point of view with Zernike moment descriptors (ZMD). These parameters are then used as inputs for a supervised classification in order to determine the most relevant. An experiment using a specific Saccharomyces cerevisiae strain presenting a fusion between a protein found in RNPs (PAB1) and the green fluorescent protein was performed to benchmark this approach. The fluorescence was observed with two-photon fluorescence microscopy. Results show that the textural approach, by mixing ZMD with Haralick features, allows for the characterization of the number of RNPs.

Faced with the massive amount of online shopping clothing images, how to recognize the attributes of clothing images quickly and accurately is a challenging task of image classification. In this paper, a new attribute classification method for clothing images is proposed based on the position relevant features acquired on the region of interest in the image. To extract the relevant regional features, one deep neural network model is designed with the pose estimation. Then the final attributes of clothing images are generated by fusing classification features. The performance of classification by fusion feature is compared with the method by single global feature of the image. The experiment results show that it is effective to combine with pose estimation in attribute classification. And the fusion feature shows good performance regardless of the neural network model.

The k-nearest neighbor (KNN) rule is a simple and effective nonparametric classification algorithm in pattern classification. However, it suffers from several problems such as sensitivity to outliers and inaccurate classification decision rule. Thus, a local mean-based k-nearest neighbor classifier (LMKNN) was proposed to address these problems, which assigns the query sample with a class label based on the closest local mean vector among all classes. It is proven that the LMKNN classifier achieves better classification performance and is more robust to outliers than the classical KNN classifier. Nonetheless, the unreliable nearest neighbor selection rule and single local mean vector strategy in LMKNN classifier severely have negative effect on its classification performance. Considering these problems in LMKNN, we propose a globally adaptive k-nearest neighbor classifier based on local mean optimization, which utilizes the globally adaptive nearest neighbor selection strategy and the implementation of local mean optimization to obtain more convincing and reliable local mean vectors. The corresponding experimental results conducted on twenty real-world datasets demonstrated that the proposed classifier achieves better classification performance and is less sensitive to the neighborhood size $$k$$ compared with other improved KNN-based classification methods.

The current detection technology for vegetable pests mainly relies on artificial statistics, which exists many shortages such as requiring a large amount of labor, low efficiency, feedback delay and artificial faults. By rapid detection and image processing technology targeting at vegetable pests, not only can reduce manpower and pesticide use, but also provide decision support for precise spraying and improve the quality of vegetables. Practical research achievements are still relatively lacking on the rapid identification technology based on image processing technology in vegetable pests. Given the above background, this paper presents a classification and recognition scheme based on the bag-of-words model and support vector machine (BOF-SVM) on four important southern vegetable pests including Whiteflies, Phyllotreta Striolata, Plutella Xylostella and Thrips. This paper consists of four sub-algorithms. The first sub-algorithm is to compute the character description of pest images based on scale-invariant feature transformation. The second sub-algorithm is to compute the visual vocabulary based on bag of features. The third sub-algorithm is to compute the classifier of pests based on support vector machines. The last one is to classify the pest images using the classifier. In this study, C++ and Python language were used as implementation technologies with OpenCV and LibSVM function library based on BOF-SVM classification algorithm. Experiments showed that the average recognition accuracy was 91.56% for a single image category judgment with 80 images from the real environment, and the average time was 0.39 seconds. This algorithm has achieved the ideal operating speed and precision. It can provide decision support for UAV precise spraying, and also has good application prospect in agriculture. Keywords: agricultural image processing, vegetable pests, classification, recognition, bag of features, support vector machine DOI: 10.25165/j.ijabe.20181103.3477 Citation: Xiao D Q, Feng J Z, Lin T Y, Pang C H, Ye Y W. Classification and recognition scheme for vegetable pests based on the BOF-SVM model. Int J Agric & Biol Eng, 2018; 11(3): 190–196.

In the last few years, bag of local features has become a popular approach for image categorization and achieved impressive performances. This approach assigns each local feature to a visual word from the visual vocabulary and represent image as a histogram of visual words. And some extensions can further improve bag of local feature approach. For example, probabilistic topic models can be adopted to capture co-occurrences between visual words in the image collection; soft vector quantization can be applied to reduce the information loss in quantization. In despite of its good performance, bag of local features approach does not correspond to human visual perception process, and studies on human perception suggest a region based approach. In this paper, we investigate applying pLSA, a probabilistic topic model, to region-based image classification, and propose two soft vector quantization methods to tackle the small sample problem in visual vocabulary construction. Our experiment indicates that applying pLSA to region-based image categorization with soft vector quantization is an effective approach.

In order to solve the issues of inadequate feature description and inefficient feature learning model existing in current classification methods, this article proposes a multi-channel joint sparse learning model for three-dimensional (3D) non-rigid object classification. First, the authors adopt a multi-level measurement of intrinsic properties to create complementary shape descriptors. Second, they build independent and informative bag of features (BoF) by embedding these shape descriptors into the visual vocabulary space. Third, a max-dependency and min-redundancy criterion is applied for optimal feature filtering on each BoF dictionary based on mutual information; meanwhile, each dictionary is learned and weighted according to its contribution to the classification task, and then a compact multi-channel joint sparse learning model is constructed. Finally, the authors train the joint sparse learning model followed by a Softmax classifier to implement efficient shape classification. The experimental results show that the proposed method has stronger feature representation ability and promotes greatly the discrimination of sparse coding coefficients. Thus, the promising classification performance and the powerful robustness can be obtained compared to the state-of-the-art methods.

ObjectivesAssessment of regional pediatric last-resort antibiotic utilization patterns is hampered by potential confounding from population differences. We developed a risk-adjustment model from readily available, internationally used survey data and a simple patient classification to aid such comparisons.DesignWe investigated the association between pediatric conserve antibiotic (pCA) exposure and patient / treatment characteristics derived from global point prevalence surveys of antibiotic prescribing, and developed a risk-adjustment model using multivariable logistic regression. The performance of a simple patient classification of groups with different expected pCA exposure levels was compared to the risk model.Setting226 centers in 41 countries across 5 continents.ParticipantsNeonatal and pediatric inpatient antibiotic prescriptions for sepsis/bloodstream infection for 1281 patients.ResultsOverall pCA exposure was high (35%), strongly associated with each variable (patient age, ward, underlying disease, community acquisition or nosocomial infection and empiric or targeted treatment), and all were included in the final risk-adjustment model. The model demonstrated good discrimination (c-statistic = 0.83) and calibration (p = 0.38). The simple classification model demonstrated similar discrimination and calibration to the risk model. The crude regional pCA exposure rates ranged from 10.3% (Africa) to 67.4% (Latin America). Risk adjustment substantially reduced the regional variation, the adjusted rates ranging from 17.1% (Africa) to 42.8% (Latin America).ConclusionsGreater comparability of pCA exposure rates can be achieved by using a few easily collected variables to produce risk-adjusted rates.