Design of a Hand Pose Recognition System for Mobile and Embedded Devices

A histogram of oriented gradient (HOG) feature is applied to the field of diseased cell detection, which can detect diseased cells in high resolution tissue images rapidly, accurately and efficiently. Firstly, motivated by symmetrical cellular forms, a new HOG symmetrical feature based on the traditional HOG feature is proposed to meet the condition of cell detection. Secondly, considering the high feature dimension of traditional HOG feature leads to plenty of memory resources and long runtime in practical applications, a classical dimension reduction method called principal component analysis (PCA) is used to reduce the dimension of high-dimensional HOG descriptor. Because of that, computational speed is increased greatly, and the accuracy of detection can be controlled in a proper range at the same time. Thirdly, support vector machine (SVM) classifier is trained with PCA-HOG symmetrical features proposed above. At last, practical tissue images is detected and analyzed by SVM classifier. In order to verify the effectiveness of this new algorithm, it is practically applied to conduct diseased cell detection which takes 200 pieces of H&E (hematoxylin & eosin) high resolution staining histopathological images collected from 20 breast cancer patients as a sample. The experiment shows that the average processing rate can be 25 frames per second and the detection accuracy can be 92.1%.

Several studies have been carried out for the rapid wood identification process without eye observation of the wood anatomists. Computer vision is the first choice in this case so that the identification results are rapid and more accurate than the conventional method. Our previous research developed a method for wood identification using the Histogram of Oriented Gradient (HOG) feature extraction and Support Vector Machine (SVM) as a classifier on Android smartphones. This paper proposes an improved wood identification accuracy of the HOG method and SVM classifier by utilizing several methods on the image preprocessing i.e. the Gaussian pyramid and the Laplacian edge detection methods. The Gaussian pyramid is used to reduce the wood image into a smaller group of pixels to qualify size wood image in the extraction process without reducing the image quality. On the other hand, to clear and distinguish the pattern in the wood image, the Laplacian edge detection is used. In our experiments, wood images from five wood species were used i.e. Kembang Semangkok, Ketapang, Preparat Darat, Pinang, and Puspa. The result showed that each wood species have increased accuracy, precision, recall, and specificity. The lowest increment accuracy was for Pinang and Puspa species at 4.00% of accuracy and zero precision value is found in Puspa species. Furthermore, from five wood species, there was a significantly increased result so it is very useful for improving the result of identification using HOG descriptor and SVM Classifier.

Communication is the ultimate of man’s search for conveying his ideas, emotions, and concepts. Dance is one of the media of communication through which dancers share notion of feelings, with the spectators through gestures, i.e., mudra. Gesture recognition propagates a concept without verbal speech or listening, and in dance recognition, the notion is transferred through various dance poses and actions. This activity in a way really paves way to enhance Indian Sign Language. This study focuses to solve the mudra resemblance in Bharatanatyam through a new system developed with image processing and classification technique using histogram of oriented gradient (HOG) feature extraction techniques and support vector machine (SVM) classifier. SVM classifies the features of HOG into mudras as text labels. Popular feature vectors such as scale-invariant feature transform (SIFT), speed up robust feature (SURF), and local binary pattern (LBP) are hardened against HOG for accuracy and speediness, and this innovative proposed concept is useful for online dance learners.

QR codes have become useful and efficient data storage tools which are exploited in many commercial applications including product tracking, website redirection, etc. A QR code is a 2-dimensional barcode localised through three finder patterns (three squares characterised by a series of alternative black and white modules at ratios 1∶1∶3∶1∶1) placed in its three corners. QR codes are generally placed in different environments with complex backgrounds (overlapping text, pictures, etc.), and are often captured under unfavourable conditions such as poor lighting. These factors can significantly affect the recognition ability and thus may hinder correct QR code localisation and identification. In order to appropriately address these issues, in this paper, we present a QR code recognition algorithm based on histogram of oriented gradients (HOG) features combined with support vector machine (SVM) classifiers. Using HOG, we extract gradient features of each extracted pattern. Subsequently, the obtained features are passed to two linear SVM classifiers, one trained with finder patterns and one trained with alignment patterns, to remove irrelevant patterns. QR codes are then conveniently localised according to a pattern closeness constraint. In the last stage, the captured code is enhanced by applying a perspective correction followed by image binarisation and morphological processing. Finally, the patterns are decoded using an accurate 2-d barcode decoder. Our proposed approach is designed for an embedded systems using a Raspberry Pi equipped with a HD camera and a small robot carrying the equipment.

This paper presents a decision method of middle ear disease which is developed in children and adults. In the proposed method, features are extracted from the middle ear disease images and normal images using HoG (histogram of oriented gradient) descriptor and the extracted features are learned by SVM (support vector machine) classifier. To obtain an input vector into SVM, an input image is resized to a predefined size and then the resized image is partitioned into 16 blocks each of which is partitioned into 4 sub-blocks (namely cell). Finally, the feature vector with 576 components is given by using HoG with 9 bins and it is used as SVM learning and classification. Input images are classified by SVM classifier based on the model of learning features. Experimental results show that the proposed method yields the precision of over 90% in decision. 키워드 : HoG, SVM, 중이염, 분류 Key word : HoG, SVM, Middle ear Disease, Classification Received 11 November 2015, Revised 14 December 2015,

This paper focuses on the accurate, combined detection of glaucoma, diabetic retinopathy, and cataracts, all using a single computer vision pipeline. Attempts have been made in past literature; however, they mainly focus on only one of the aforementioned eye diseases. These diseases must be identified in the early stages to prevent damage progression. Three pipelines were constructed, of which 12 deep learning models and 8 Support Vector Machines (SVM) classifiers were trained. Pipeline 1 extracted Histogram of Oriented Gradients (HOG) features, and pipeline 2 extracted Grey-Level Co-occurrence Matrix (GLCM) textural features from the pre-processed images. These features were classified with either a linear or Radial Basis Function (RBF) kernel SVM. Pipeline 3 utilised various deep learning architectures for feature extraction and classification. Two models were trained for each deep learning architecture and SVM classifier, using standard RGB images (labelled as Normal). The other uses retina images with only the green channel present (labelled as Green). The Inception V3 Normal model achieved the best performance with accuracy and an F1-Score of 99.39%. The SqueezeNet Green model was the worst-performing deep learning model with accuracy and an F1-Score of 81.36% and 81.29%, respectively. Although it performed the worst, the model size is 5.03 MB compared to the 225 MB model size of the top-performing Inception V3 model. A GLCM feature selection study was performed for both the linear and RBF SVM kernels. The RBF SVM that extracted HOG features on the green-channel images performed the best out of the SVMs with accuracy and F1-Score of 76.67% and 76.48%, respectively. The green-channel extraction was more effective on the SVM classifiers than the deep learning models. The Inception V3 Normal model can be integrated with a computer-aided system to facilitate examiners in detecting diabetic retinopathy, cataracts and glaucoma.KeywordsGlaucomaDiabetic retinopathyCataractComputer visionConvolutional neural networkDeep learningGLCMHOGCAD

Human action recognition from 2D videos is a demanding area due to its broad applications. Many methods have been proposed by the researchers for recognizing human actions. The improved accuracy in identifying human actions is desirable. This paper presents an improved method of human action recognition using support vector machine (SVM) classifier. This paper proposes a novel feature descriptor constructed by fusing the various investigated features. The handcrafted features such as scale invariant feature transform (SIFT) features, speed up robust features (SURF), histogram of oriented gradient (HOG) features and local binary pattern (LBP) features are obtained on online 2D action videos. The proposed method is tested on different action datasets having both static and dynamically varying backgrounds. The proposed method achieves shows best recognition rates on both static and dynamically varying backgrounds. The datasets considered for the experimentation are KTH, Weizmann, UCF101, UCF sports actions, MSR action and HMDB51.The performance of the proposed feature fusion model with SVM classifier is compared with the individual features with SVM. The fusion method showed best results. The efficiency of the classifier is also tested by comparing with the other state of the art classifiers such as k-nearest neighbors (KNN), artificial neural network (ANN) and Adaboost classifier. The method achieved an average of 94.41% recognition rate.

Recently, in order to solve the problem of image classification, some image features and classifiers play more and more important role in the related research field. This article investigates an image classification method by the histogram of oriented gradient (HOG) features, the gray-level co-occurrence matrix (GLCM) features, and the support vector machine (SVM) classifier. By obtaining the HOG and the GLCM features of image, the combination of them is inputted into the SVM for the training and the test. The experiment results have manifested the effectiveness of the proposed method. The use of the combination of HOG features and GLCM features in image classification is far superior to the use of them alone.

Support Vector Machine (SVM) classifier with Histogram of Orientated Gradients (HOG) feature is one of the most popular techniques used for vehicle detection in recent years. In this paper, we study the effect of HOG parameter values on the performance and computing time of vehicle detection. The aim of this paper is to explore the relationship between performance/computing time and HOG parameter values, and eventually to guide finding the most appropriate parameter set to meet specific problem constrains.

Forest areas in Indonesia covered about 2/3 of total land areas which has about 4000 wood species. Wood identification plays a key role in wood utilization not only for determining appropriate use but also for supporting legal timber trade. However, the identification process requires high expertise and complex method which can be done in the laboratory. In order to simplify the identification process, we develop wood identification using computer vision by using Histogram of Oriented Gradient (HOG) to extract the species of wood and Support Vector Machines (SVM) to classify wood species. These methods combination will improve the accuracy of wood identification process. The result showed that the HOG method can extract the texture of woods and SVM classifier can generate the boundary decision after executing the training process. By doing the testing process of SVM classifier, the result showed that the accuracy from the identification is 70.5% for using positive testing image and 77.5% for using negative testing image. This accuracy value can be reached because the texture for each training image has different texture pattern especially the number and location of vessels.

Computer vision is an important sensing technique to translate the information to decisions. In robotic applications, object detection is a critical skill to perform tasks for robots in complex environments. The deep-learning framework, e.g. You Only Look Once (YOLO), attracts much more attention recently. Moreover, it is not an optimal solution for a mobile robot since it requires a large scale of hardware resources, on-chip SRAMs, and power consumption. In this work, we report an object detection processor synchronizing the image sensor in FPGA with a cellbased histogram of oriented gradient (HOG) feature descriptor and support vector machine (SVM) classifier by parallel sliding window mechanism. The HOG feature extraction circuitry with pixel-based pipelined architecture constructs the cell-based feature vectors for parallelizing partial SVM-based classification in multiple sliding windows. The SVM classification produces the final result after accumulating the vector components in one sliding window. This framework can be used to both localize and recognize multiple objects in video footage. The proposed object processor, in which the SVM classifier is trained by INRIA datasets, is implemented and verified on Intel Stratix IV FPGA for the pedestrian.

Hardware implementation of human detection is a challenging task for embedded designs. This paper presents a real-time image-based field-programmable gate array (FPGA) implementation of human detection. Our implementation is based on the histograms of oriented gradients (HOG) feature and linear support vector machine (SVM) classifier. The novelty of this work is that we replace normalization process of HOG with a modified binarization process. Therefore, during classification process with SVM classifier, all multiplication operations are replaced by addition operations. All these modifications result in reduction of hardware resource. Experimental evaluation reveals that 293 fps can be achieved on a low-end Xilinx Spartan-3e FPGA. Moreover, a detection accuracy of 1.97% miss rate and 1% false positive rate is achieved. For further demonstration, a prototype system is developed with an OV7670 camera device. Restricted to the speed of camera, a detection rate of 30 fps is achieved.

Facial Expression Recognition (FER) has been an active topic of papers that were researched during 1990s till now, according to its importance, FER has achieved an extremely role in image processing area. FER typically performed in three stages include, face detection, feature extraction and classification. This paper presents an automatic system of face expression recognition which is able to recognize all eight basic facial expressions which are (normal, happy, angry, contempt, surprise, sad, fear and disgust) while many FER systems were proposed for recognizing only some of face expressions. For validating the method, the Extended Cohn-Kanade (CK+) dataset is used. The presented method uses Viola-Jones algorithm for face detection. Histogram of Oriented Gradients (HOG) is used as a descriptor for feature extraction from the images of expressive faces. Principal Component Analysis (PCA) applied to reduce dimensionality of the Features, to obtaining the most significant features. Finally, the presented method used three different classifiers which are Support Vector Machine (SVM), K-Nearest Neighbor (KNN) and Multilayer Perceptron Neural Network (MLPNN) for classifying the facial expressions and the results of them are compared. The experimental results show that the presented method provides the recognition rate with 93.53% when using SVM classifier, 82.97% when using MLP classifier and 79.97% when using KNN classifier which refers that the presented method provides better results while using SVM as a classifier.

One of the major agricultural disasters, which depressed crop production and lower-quality, is pest in China. The lack of technical and scientific knowledge to prevent pest diseases is the main reason for low productivity of these crop commodities. Traditional methods based on artificial judgment have the disadvantages of large workload, low efficiency, poor working environment, and caused damage to crops. In response to the above problems, a HOG + SVM-based crop pest monitoring drone (PM-VAV) system was proposed, which utilizes computer vision combined with the flexibility of the drone for non-contact measurement. Firstly, the real aircraft working platform of this PM-UAV system was constructed. Then, through self-built data sets satisfy the requirements of the model training for pest detection task. Secondly, the on-line monitoring task of crop pests was accomplished through the use of the Histogram of Oriented Gradient (HOG) feature and Support Vector Machine (SVM) classifier, the algorithm was ported to the airborne embedded platform NVIDIA TK1. Finally, experimental tests show that the designed monitoring aircraft can effectively implement on-line monitoring for the crop pest which contained in the self-built data set.

This paper presents the used of histogram of oriented gradient (HOG) for facial expression recognition using support vector machine (SVM). In this work, the facial expression images are firstly preprocessed by face detection and cropped images. Then, HOG method is adopted as feature extraction on facial image. The ability of HOG to preserve the local information and orientation density distribution in facial images suitable as shape descriptor for facial expression. It divides the image into cell or patch that has magnitude and orientations. The extracted HOG was then concatenated into histogram bin to form one feature vector before feed into SVM classifier. Both JAFFE and KDEF datasets were employed to evaluate the performance of proposed method. Based on results, the average recognition rates of JAFFE and KDEF datasets are 76.19% and 80.95% respectively. The results show that the performance of expression surprise has outperformed compared to others expression while expression fear contributes the lowest recognition rate. Thus, utilization of HOG features with SVM classifier have shown the promising results in recognizing facial expression.