Search results

Single-shot multi-category clothing recognition and retrieval play a crucial role in online searching and offline settlement scenarios. Existing clothing recognition methods based on RGBD clothing images often suffer from high-dimensional feature representations, leading to compromised performance and efficiency.

To address this issue, this paper proposes a novel method called Manifold Embedded Discriminative Feature Selection (MEDFS) to select global and local features, thereby reducing the dimensionality of the feature representation and improving performance. Specifically, by combining three global features and three local features, a low-dimensional embedding is constructed to capture the correlations between features and categories. The MEDFS method designs an optimization framework utilizing manifold mapping and sparse regularization to achieve feature selection. The optimization objective is solved using an alternating iterative strategy, ensuring convergence.

Empirical studies conducted on a publicly available RGBD clothing image dataset demonstrate that the proposed MEDFS method achieves highly competitive clothing classification performance while maintaining efficiency in clothing recognition and retrieval.

This paper introduces a novel approach for multi-category clothing recognition and retrieval, incorporating the selection of global and local features. The proposed method holds potential for practical applications in real-world clothing scenarios.

Displacement measurement in large-scale structures (such as excavation walls) is one of the most important applications of close-range photogrammetry, in which achieving high precision requires extracting and accurately matching local features from convergent images. The purpose of this study is to introduce a new multi-image pointing (MIP) algorithm is introduced based on the characteristics of the geometric model generated from the initial matching. This self-adaptive algorithm is used to correct and improve the accuracy of the extracted positions from local features in the convergent images.

In this paper, the new MIP algorithm based on the geometric characteristics of the model generated from the initial matching was introduced, which in a self-adaptive way corrected the extracted image coordinates. The unique characteristics of this proposed algorithm were that the position correction was accomplished with the help of continuous interaction between the 3D model coordinates and the image coordinates and that it had the least dependency on the geometric and radiometric nature of the images. After the initial feature extraction and implementation of the MIP algorithm, the image coordinates were ready for use in the displacement measurement process. The combined photogrammetry displacement adjustment (CPDA) algorithm was used for displacement measurement between two epochs. Micro-geodesy, target-based photogrammetry and the proposed MIP methods were used in a displacement measurement project for an excavation wall in the Velenjak area in Tehran, Iran, to evaluate the proposed algorithm performance. According to the results, the measurement accuracy of the point geo-coordinates of 8 mm and the displacement accuracy of 13 mm could be achieved using the MIP algorithm. In addition to the micro-geodesy method, the accuracy of the results was matched by the cracks created behind the project’s wall. Given the maximum allowable displacement limit of 4 cm in this project, the use of the MIP algorithm produced the required accuracy to determine the critical displacement in the project.

Evaluation of the results demonstrated that the accuracy of 8 mm in determining the position of the points on the feature and the accuracy of 13 mm in the displacement measurement of the excavation walls could be achieved using precise positioning of local features on images using the MIP algorithm.The proposed algorithm can be used in all applications that need to achieve high accuracy in determining the 3D coordinates of local features in close-range photogrammetry.

Some advantages of the proposed MIP photogrammetry algorithm, including the ease of obtaining observations and using local features on the structure in the images rather than installing the artificial targets, make it possible to effectively replace micro-geodesy and instrumentation methods. In addition, the proposed MIP method is superior to the target-based photogrammetric method because it does not need artificial target installation and protection. Moreover, in each photogrammetric application that needs to determine the exact point coordinates on the feature, the proposed algorithm can be very effective in providing the possibility to achieve the required accuracy according to the desired objectives.

In order to solve the problem that the performance of the existing local feature descriptors in uncontrolled environment is greatly affected by illumination, background, occlusion and other factors, we propose a novel face recognition algorithm in uncontrolled environment which combines the block central symmetry local binary pattern (CS-LBP) and deep residual network (DRN) model.

The algorithm first extracts the block CSP-LBP features of the face image, then incorporates the extracted features into the DRN model, and gives the face recognition results by using a well-trained DRN model. The features obtained by the proposed algorithm have the characteristics of both local texture features and deep features that robust to illumination.

Compared with the direct usage of the original image, the usage of local texture features of the image as the input of DRN model significantly improves the computation efficiency. Experimental results on the face datasets of FERET, YALE-B and CMU-PIE have shown that the recognition rate of the proposed algorithm is significantly higher than that of other compared algorithms.

The proposed algorithm fundamentally solves the problem of face identity recognition in uncontrolled environment, and it is particularly robust to the change of illumination, which proves its superiority.

An accurate detection of overhead ground wire under open surroundings with varying illumination is the premise of reliable line grasping with the off-line arm when the inspection robot cross obstacle automatically. This paper aims to propose an improved approach which is called adaptive homomorphic filter and supervised learning (AHSL) for overhead ground wire detection.

First, to decrease the influence of the varying illumination caused by the open work environment of the inspection robot, the adaptive homomorphic filter is introduced to compensation the changing illumination. Second, to represent ground wire more effectively and to extract more powerful and discriminative information for building a binary classifier, the global and local features fusion method followed by supervised learning method support vector machine is proposed.

Experiment results on two self-built testing data sets A and B which contain relative older ground wires and relative newer ground wire and on the field ground wires show that the use of the adaptive homomorphic filter and global and local feature fusion method can improve the detection accuracy of the ground wire effectively. The result of the proposed method lays a solid foundation for inspection robot grasping the ground wire by visual servo.

This method AHSL has achieved 80.8 per cent detection accuracy on data set A which contains relative older ground wires and 85.3 per cent detection accuracy on data set B which contains relative newer ground wires, and the field experiment shows that the robot can detect the ground wire accurately. The performance achieved by proposed method is the state of the art under open environment with varying illumination.

Clothing retrieval is very useful to help the clients to efficiently search out the apparel they want. Currently, the mainstream clothing retrieval methods are attribute semantics based, which however are inconvenient for common clients. The purpose of this paper is to provide an easy‐to‐operate apparels retrieval mode with the authors' novel approach of clothing image similarity measurement.

The authors measure the similarity between two clothing images by computing the weighted similarities between their bundled features. Each bundled feature consists of the point features (SIFT) which are further quantified into local visual words in a maximally stable extremal region (MSER). The authors weight the importance of bundled features by the precision of SIFT quantification and local word frequency that reflects the frequency of the common visual words appeared in two bundled features. The bundled features similarity is computed from two aspects: local word frequency; and SIFTs distance matrix that records the distances between every two SIFTs in a bundled feature.

Local word frequencies improves the recognition between two bundled features with the same common visual words but different local word frequency. SIFTs distance matrix has the merits of scale invariance and rotation invariance. Experimental results show that this approach works well in the situations with large clothing deformation, background exchange and part hidden, etc. And the similarity measurement of Weight+Bundled+LWF+SDM is the best.

This paper presents an apparel retrieval mode based on local visual features, and presents a new algorithm for bundled feature matching and apparel similarity measurement.

Image classification or more specifically, annotating images with keywords is one of the important steps during image database indexing. However, the problem with current research in terms of image retrieval is more concentrated on how conceptual categories can be well represented by extracted, low level features for an effective classification. Consequently, image features representation including segmentation and low‐level feature extraction schemes must be genuinely effective to facilitate the process of classification. The purpose of this paper is to examine the effect on annotation effectiveness of using different (local) feature representation methods to map into conceptual categories.

This paper compares tiling (five and nine tiles) and regioning (five and nine regions) segmentation schemes and the extraction of combinations of color, texture, and edge features in terms of the effectiveness of a particular benchmark, automatic image annotation set up. Differences between effectiveness on concrete or abstract conceptual categories or keywords are further investigated, and progress towards establishing a particular benchmark approach is also reported.

In the context of local feature representation, the paper concludes that the combined color and texture features are the best to use for the five tiling and regioning schemes, and this evidence would form a good benchmark for future studies. Another interesting finding (but perhaps not surprising) is that when the number of concrete and abstract keywords increases or it is large (e.g. 100), abstract keywords are more difficult to assign correctly than the concrete ones.

Future work could consider: conduct user‐centered evaluation instead of evaluation only by some chosen ground truth dataset, such as Corel, since this might impact effectiveness results; use of different numbers of categories for scalability analysis of image annotation as well as larger numbers of training and testing examples; use of Principle Component Analysis or Independent Component Analysis, or indeed machine learning techniques for low‐level feature selection; use of other segmentation schemes, especially more complex tiling schemes and other regioning schemes; use of different datasets, use of other low‐level features and/or combination of them; use of other machine learning techniques.

This paper is a good start for analyzing the mapping between some feature representation methods and various conceptual categories for future image annotation research.

To realize stable and precise localization in the dynamic environments, the authors propose a fast and robust visual odometry (VO) approach with a low-cost Inertial Measurement Unit (IMU) in this study.

The proposed VO incorporates the direct method with the indirect method to track the features and to optimize the camera pose. It initializes the positions of tracked pixels with the IMU information. Besides, the tracked pixels are refined by minimizing the photometric errors. Due to the small convergence radius of the indirect method, the dynamic pixels are rejected. Subsequently, the camera pose is optimized by minimizing the reprojection errors. The frames with little dynamic information are selected to create keyframes. Finally, the local bundle adjustment is performed to refine the poses of the keyframes and the positions of 3-D points.

The proposed VO approach is evaluated experimentally in dynamic environments with various motion types, suggesting that the proposed approach achieves more accurate and stable location than the conventional approach. Moreover, the proposed VO approach works well in the environments with the motion blur.

The proposed approach fuses the indirect method and the direct method with the IMU information, which improves the localization in dynamic environments significantly.

Diagnosing pain in neonates is difficult but critical. Although approximately thirty manual pain instruments have been developed for neonatal pain diagnosis, most are complex, multifactorial, and geared toward research. The goals of this work are twofold: 1) to develop a new video dataset for automatic neonatal pain detection called iCOPEvid (infant Classification Of Pain Expressions videos), and 2) to present a classification system that sets a challenging comparison performance on this dataset. The iCOPEvid dataset contains 234 videos of 49 neonates experiencing a set of noxious stimuli, a period of rest, and an acute pain stimulus. From these videos 20 s segments are extracted and grouped into two classes: pain (49) and nopain (185), with the nopain video segments handpicked to produce a highly challenging dataset. An ensemble of twelve global and local descriptors with a Bag-of-Features approach is utilized to improve the performance of some new descriptors based on Gaussian of Local Descriptors (GOLD). The basic classifier used in the ensembles is the Support Vector Machine, and decisions are combined by sum rule. These results are compared with standard methods, some deep learning approaches, and 185 human assessments. Our best machine learning methods are shown to outperform the human judges.

Palmprint recognition is a very interesting and promising area of research. Much work has already been done in this area, but much more needs to be done to make the systems more efficient. In this paper, a multimodal biometrics system based on fusion of left and right palmprints of a person is proposed to overcome limitations of unimodal systems.

Features are extracted using some proposed multi-block local descriptors in addition to MBLBP. Fusion of extracted features is done at feature level by a simple concatenation of feature vectors. Then, feature selection is performed on the resulting global feature vector using evolutionary algorithms such as genetic algorithms and backtracking search algorithm for a comparison purpose. The benefits of such step selecting the relevant features are known in the literature, such as increasing the recognition accuracy and reducing the feature set size, which results in runtime saving. In matching step, Chi-square similarity measure is used.

The resulting feature vector length representing a person is compact and the runtime is reduced.

Intensive experiments were done on the publicly available IITD database. Experimental results show a recognition accuracy of 99.17 which prove the effectiveness and robustness of the proposed multimodal biometrics system than other unimodal and multimodal biometrics systems.

The purpose of this paper is to build a mobile visual search service system for the protection of Dunhuang cultural heritage in the smart library. A novel mobile visual search model for Dunhuang murals is proposed to help users acquire rich knowledge and services conveniently.

First, local and global features of images are extracted, and the visual dictionary is generated by the k-means clustering. Second, the mobile visual search model based on the bag-of-words (BOW) and multiple semantic associations is constructed. Third, the mobile visual search service system of the smart library is designed in the cloud environment. Furthermore, Dunhuang mural images are collected to verify this model.

The findings reveal that the BOW_SIFT_HSV_MSA model has better search performance for Dunhuang mural images when the scale-invariant feature transform (SIFT) and the hue, saturation and value (HSV) are used to extract local and global features of the images. Compared with different methods, this model is the most effective way to search images with the semantic association in the topic, time and space dimensions.

Dunhuang mural image set is a part of the vast resources stored in the smart library, and the fine-grained semantic labels could be applied to meet diverse search needs.

The mobile visual search service system is constructed to provide users with Dunhuang cultural services in the smart library. A novel mobile visual search model based on BOW and multiple semantic associations is proposed. This study can also provide references for the protection and utilization of other cultural heritages.