Search

Deaths and injuries resulting from road traffic crashes remain a serious problem globally and current trends suggest that this will continue to be the case in the foreseeable future. In this paper, we propose a reliable method towards drowsiness detection by analyzing images of the driver’s face. In fact, the shadows caused by wearing glasses and/or bad light conditions may, in particular, decrease the accuracy rate of the blinking detection. In addition, drowsiness symptoms are not just restricted to the frequency of blinking, as most of existing works do, but also include yawning and nodding. Thus, instead of using a single facial feature to predict the drowsiness state, we jointly combine eye closure and yawning by measuring the eye and the mouth aspect ratios, and head pose which is estimated by analyzing the optical flow. Then, we investigate the multilayer perceptron and the K-nearest neighbors as two classification techniques. The proposed method has been tested on the challenging private NTHU-DDD benchmark video dataset, and the preliminary experimental results show the effectiveness of the proposed automated method.

Inspection has been always a topic of interest in the industrial environment. Indeed, the inspection process is particularly very essential for controlling the quality of the finished products. Within this context, in this work is proposed an automated artificial vision system in order to control the non-conformities in vehicles. In fact, the main idea of the proposed system resides in the adoption of the Artificial Intelligence (AI), in order to identify automatically the defects, while supporting at the same time the concept of a smart industry 4.0. Moreover, some works, which rely on AI for quality control inspection in smart industries, are briefly reviewed in this paper, in order to highlight the issues of Visual Inspection (VI) as well as to prove the effectiveness of the AI in industries, notably those related to vehicle manufacturing.

This research presents an automatic method to detect and evaluate the shoreline changes from Landsat satellite images. In fact, a method, that we called Lukas-Kanade Adapted for Coastal Changes (LKA2C), has been developed to calculate and detect the changes around the study region. Mainly the proposed method is based on SURF algorithm to detect the study region from the satellite image. Then, Canny edge detector was used on NDWI images to detect the shorelines. Finally, the pyramidal Lukas-Kanade optical flow algorithm was adapted to detect and to calculate the rates of changes. Realized experiments on real satellite images of the island of Djerba in Tunisia proved the effectiveness of the proposed method.

Semantic gap, which is the difference between low-level image features and their high-level semantics, has become very popular and witnessed great interest in the last two decades. This paper deals with this problem and proposes a hybrid approach to learn image semantic concepts for modeling visual features in discriminative learning stage. It combines the advantages of human-in-the-loop and discriminative semantic models. Herein, we investigate the expert-domain knowledge and expertise owing to expert-in-the-loop to determine medical-knowledge informations. Semantic models aim to learn the correlations between low-level features and textual words to describe malignancy signs in terms of semantic visual descriptors. These descriptors are automatically generated from low-level image features by exploiting the semantic concepts-based clinician medical-knowledge. Reported results over mammography image analysis society (MIAS) database prove the effectiveness of this work and its outperformance relative to compared approaches.

Content-based image retrieval (CBIR) is a primordial task to provide the most similar images especially in the context of medical imaging for diagnosis aid. In this paper, we propose a CBIR method for a large-scale mammogram datasets. In fact, to extract region of interest (ROI) signatures, four moment descriptors were defined after computing the curvelet coefficients for each level of the ROI. Then, an unsupervised technique based on locality sensitive hashing was adopted for indexing the extracted signatures. The main contribution of the suggested method resides in the variance-based filtering within the retrieval phase in order to extract the suitable buckets in the shortest time, while optimizing the memory requirement. After that, an accurate searching in Hamming space is performed in order to identify the similar ROIs to the query case. Realized experiments on the challenging Digital Database for Screening Mammography (DDSM) dataset proved the performance of the proposed method for the retrieval of the most relevant mammograms in a large-scale dataset. It achieves a mean retrieval precision rate of 97.1% over a total of 11218 mammogram ROIs.

Retrieval engines provide results according to user request. Nevertheless, reaching satisfaction can not be guaranteed with simple retrieval step. Therefore, it is necessary to communicate this dissatisfaction to the system through relevance feedback techniques. Indeed, with the growing number of image collections and by applying approximate nearest neighbor (ANN) algorithms to resolve the curse of dimensionality, the semantic gap may increase. For this reason, an additional step of relevance feedback is needful to add semantics to the next retrieval iterations. In this paper, a classification of the different relevance feedback techniques related to region-based image retrieval applications is elaborated. Moreover, a new technique of relevance feedback based on re-weighting regions of the query-image by selecting a set of negative examples is detailed. Furthermore, the general context to carry out this technique which is the large-scale heterogeneous image collections indexing and retrieval is presented. In fact, the main contribution of the proposed work is affording efficient results with the minimum number of relevance feedback iterations for high dimensional image databases. Experiments and assessments are carried out within an RBIR system for "Wang" data set in order to prove the effectiveness of the proposed approaches.

In this paper we are interested in the semi-supervised image classification in large datasets. The main originality of the proposed technique resides in the fuzzy quantification of the salient object in each image in order to guide the semi-supervised learning process during the classification. Indeed, we detect the salient object in each image using soft image abstraction, which allows the subsequent global saliency cues to uniformly highlight entire salient regions. Then, fuzzy quantification was involved for the purpose of improving the correct belonging of pixels to the salient object in each image. For classification, ensemble projection is used, while training a random forest classifier on labeled images with the learned features to classify the unlabeled ones. Experimental results on two challenging large benchmarks show the accuracy and the efficiency of the proposed technique.

Images comparison is the most critical step in the content-based image retrieval process. Therefore, we propose in this paper our approach of comparison based on image modeling by complete oriented graph. This structure encompasses low-level region descriptors in nodes and coarse spatial disposition in edges. Each node is characterized by its wavelet transformation high frequency sub-band weighted by the region importance. Similarity degree between two images is identified thereafter by comparing their graphs using heuristics to guarantee low computational overhead and to resolve the NP-hard matching problem between graphs. The experimental results and comparison made with similar image retrieval engines indicate the robustness of the proposed approach for Wang dataset and prove the applied heuristics.

Avoiding the “curse of dimensionality” in contentbased image retrieval becomes one of the most essential tasks to achieve because of the high number of stocked images as well as the high dimensionality of the descriptive vectors' space. In this context, our work consists on minimizing low-level features describing an image by using a reduced descriptor that combines color and texture information which is wavelet transformation. In fact, we propose to describe the image by high frequency subbands of discrete wavelet transformation (DWT) related to weighted salient regions after a fuzzy segmentation step. Moreover, images comparison guided by the most weighted regions is presented. Experiments and comparative study with other similar works prove the efficiency of the proposed approach for image retrieval in heterogeneous image bases.