The appearance of a face image is severely affected by illumination conditions that hinder the automatic recognition process. To recognize faces under varying illuminations, we propose wavelet-based normalization method so as to normalize illuminations. This enhances contrast well edges images simultaneously, in frequency domain using wavelet transform, facilitate tasks. It outperforms conventional - histogram equalization only pixel gray-level spatial domain. With this method, our system...

Abstract. Particle sensing technology has shown great potential for monitoring particulate matter (PM) with very few temporal and spatial restrictions because of its low cost, compact size, easy operation. However, the performance low-cost sensors PM in ambient conditions not been thoroughly evaluated. Monitoring results by are often questionable. In this study, a fine particle monitor (Plantower PMS 5003) was colocated reference instrument, Synchronized Hybrid Ambient Real-time Particulate...

We present an algorithm for blindly recovering constituent source spectra from magnetic resonance (MR) chemical shift imaging (CSI) of the human brain. The algorithm, which we call constrained nonnegative matrix factorization (cNMF), does not enforce independence or sparsity, instead only requiring and mixing matrices to be nonnegative. It is based on (NMF) extending it include a constraint positivity amplitudes recovered spectra. This enables recovery physically meaningful even in presence...

Variable illumination conditions, especially the side lighting effects in face images, form a main obstacle recognition systems. To deal with this problem, paper presents novel adaptive region-based image preprocessing scheme that enhances images and facilitates invariant task. The proposed method first segments an into different regions according to its local then both contrast edges are enhanced regionally so as alleviate effect. Different from existing enhancement methods, we apply...

Manual grading of essays by humans is time-consuming and likely to be susceptible inconsistencies inaccuracies. In recent years, an abundance research has been done automate essay evaluation processes, yet little take into consideration the syntax, semantic coherence sentiments essay's text together. Our proposed system incorporates not just rule-based grammar surface level check but also includes similarity sentences. We propose use Graph-based relationships within content polarity opinion...

Gas leakage poses a significant hazard that requires prevention. Traditionally, human inspection has been used for detection, slow and labour-intensive process. Recent research applied machine learning techniques to this problem, yet there remains shortage of high-quality, publicly available datasets. This paper introduces synthetic dataset featuring diverse backgrounds, interfering foreground objects, leak locations, precise segmentation ground truth. We propose zero-shot method combines...

Human's safety in construction areas is vital. A hard hat required to enter a area. Stopping person who not wearing entering area very important. Video-based surveillance detect new solution this problem. This paper brings different video processing techniques together construct framework for fast and robust detection areas. The proposed system can face real time.

We propose a polarization-based probabilistic discriminative model for deriving set of new sigmoid-transformed polarimetry feature parameters, which not only enables accurate and quantitative characterization cancer cells at pixel level, but also accomplish the task with simple stable model. By taking advantages polarization imaging techniques, these parameters enable low-magnification wide-field system to separate types into more specific categories that previously were distinctive under...

To securely transmit secret data between Internet of Things (IoT) nodes, it is required to the implement information hiding technique for secure communication in IoT environment. The traditional approaches generally select a multimedia file, such as texts, images, and video clips cover, then embed into cover by slight modification. However, not feasible directly apply these environment following reasons. First, hard some nodes effectively efficiently process complex data. Second,...

Polarization imaging is sensitive to sub-wavelength microstructures of various cancer tissues, providing abundant optical characteristics and microstructure information complex pathological specimens. However, how reasonably utilize polarization strengthen diagnosis ability remains a challenging issue. In order take full advantage image features samples, we propose dual modality fusion network (DPMFNet), which consists multi-stream CNN structure switched attention module for complementarily...

The pose variation involved in facial images significantly degrades the performance of face recognition systems. In this paper, a component-wise normalization method for facilitating pose-invariant is proposed. main idea to normalize non-frontal image virtual frontal component by component. method, we first partition whole into different components and then view each estimated separately. final generated integrating components. proposed relies only on 2D images, therefore complex 3D modeling...

As a powerful global registration method for point clouds, the 4-points congruent sets (4PCS) algorithm has been wildly used in 3D scene reconstruction field. In this paper, we propose an adaptive 4PCS (A4PCS), which aims to provide robustness rigid transformation two or more overlapping laser scans. The proposed only incorporates distance information of stereoscopic base set and fast mechanism extraction into 4PCS. To ensure accuracy when dealing with restrictive situations, such as clouds...

A leak detection and repair survey (LDAR) is essential to ensure a reliable safe liquefied natural gas (LNG) supply. Modern LDAR systems deploy numerous fixed thermal imaging devices automatically monitor the risk of potential leaks empowered by computational intelligence frameworks. Existing frameworks employ either background subtraction-based (BGS-based) or deep neural network-based (DNN-based) for LNG from images. However, BGS-based feature high sensitivity perceive emissions with low...

In recent years, many fusion algorithms based on multi-scale transform or neural networks have been proposed to improve medical image (MIF) performance. However, there is still enormous potential explore the combination of different theories. this paper, we propose a novel MIF framework integrate powerful feature representation abilities deep learning model and accurate frequency decomposition characteristics discrete wavelet (DWT). Firstly, encoder-decoder network well-trained extract...

We present algorithms for planning dynamically feasible backward paths and controlling unmanned surface vehicles along such using a nonlinear backstepping trajectory tracking controller. These have been developed Virginia Tech's riverine vehicle (USV). The USV is rigid hull inflatable boat with servo-actuated outboard motor. path algorithm uses the receding horizon approach solves an optimal control problem by converting it into programming (NLP) problem. law modification of forward law,...

There is an increasing demand from biosciences to develop new and efficient techniques assist in the preparation analysis of biological samples such as cells suspension. A dielectrophoresis (DEP)-based characterization measurement technique on opens up a broader perspective for early diagnosis diseases. An real-time multiple-cell tracking platform coupled with DEP capture quantify dynamics cell motion obtain viability information presented. The procedure single DEP-levitated Canola plant...