Varying Illumination Conditions Research Articles

Recognizing faces prone to occlusions and common variations using optimal face subgraphs

An intuitive graph optimization face recognition approach called Harmony Search Oriented-EBGM (HSO-EBGM) inspired by the classical Elastic Bunch Graph Matching (EBGM) graphical model is proposed in this contribution. In the proposed HSO-EBGM, a recent evolutionary approach called harmony search optimization is tailored to automatically determine optimal facial landmarks. A novel notion of face subgraphs have been formulated with the aid of these automated landmarks that maximizes the similarity entailed by the subgraphs. For experimental evaluation, two sets of de facto databases (i.e., AR and Face Recognition Grand Challenge (FRGC) ver2.0) are used to validate and analyze the behavior of the proposed HSO-EBGM in terms of number of subgraphs, varying occlusion sizes, face images under controlled/ideal conditions, realistic partial occlusions, expression variations and varying illumination conditions. For a number of experiments, results justify that the HSO-EBGM shows improved recognition performance when compared to recent state-of-the-art face recognition approaches.

Effects of the partitioning of diffuse and direct solar radiation on satellite-based modeling of crop gross primary production

Modeling crop gross primary production (GPP) is critical to understanding the carbon dynamics of agro-ecosystems. Satellite-based studies have widely used production efficiency models (PEM) to estimate cropland GPP, wherein light use efficiency (LUE) is a key model parameter. One factor that has not been well considered in many PEMs is that canopy LUE could vary with illumination conditions. This study investigates how the partitioning of diffuse and direct solar radiation influences cropland GPP using both flux tower and satellite data. The field-measured hourly LUE under cloudy conditions was 1.50 and 1.70 times higher than that under near clear-sky conditions for irrigated corn and soybean, respectively. We applied a two-leaf model to simulate the canopy radiative transfer process, where modeled photosynthetically active radiation (PAR) absorbed by canopy agreed with tower measurements (R2=0.959 and 0.914 for corn and soybean, respectively). Derived canopy LUE became similar after accounting for the impact of light saturation on leaf photosynthetic capacity under varied illumination conditions. The impacts of solar radiation partitioning on satellite-based modeling of crop GPP was examined using vegetation indices (VI) derived from MODIS data. Consistent with the field modeling results, the relationship between daily GPP and PAR×VI under varied illumination conditions showed different patterns in terms of regression slope and intercept. We proposed a function to correct the influences of direct and diffuse radiation partitioning and the explained variance of flux tower GPP increased in all experiments. Our results suggest that the non-linear response of leaf photosynthesis to light absorption contributes to higher canopy LUE on cloudy days than on clear days. We conclude that accounting for the impacts of solar radiation partitioning is necessary for modeling crop GPP on a daily or shorter basis.

Adaptive thresholding with fusion using a RGBD sensor for red sweet-pepper detection

An adaptive thresholding algorithm combined with sensor fusion to detect sweet-peppers with high detection rates for highly variable illumination conditions is presented. The objectives were to develop an algorithm for sweet-pepper detection, robust to illumination changes, with the aim of achieving high detection rates with low false alarms, that was robust to the selected thresholds and noise coming from the camera itself. An algorithm was developed that firstly divided the image into rectangular-shaped sub-images with approximately homogenous lighting conditions. The image was then further divided into amorphous-shaped sub-images and the RGB image was transformed to a 3D natural difference index image; for each sub-image, three thresholds were adaptively calculated and applied. In a final step, morphological operations and fusion with a depth sensor were developed to reduce false positives. Intensive evaluation was conducted on a database with normal lighting conditions. Twenty-five images were found to be the minimum training set required to achieve best performances. The true-positive rate was 0.909 and the false-positive rate was 0.046. Added noise up to 2% from the maximum available scale was found to have almost no influence on the performance. Validation using a second database with artificial lighting resulted with a true-positive rate of 1.00 and false-positive rate of 0.061, which suggests the importance of illumination. Fusion was critical for increasing the true-positive rate. The algorithm can also be used for detection in other crops.

A Non Invasive Heart Rate Measurement System for Multiple People in the Presence of Motion and Varying Illumination

A real time non-invasive heart rate (HR) measurement system for multiple people in an image has been proposed in this paper. The data has been gathered using a webcam with different distances with the subjects under varying illumination conditions. The effect of motion of the subjects on HR measurement is also observed. The data is gathered from the face and cheeks are selected as the region of interest (ROI). RGB color model is used for processing. Fast Fourier transform is used to detect the peak frequency after band pass filtering. The green channel of RGB color model gives best HR measurement. The minimum percentage error of 3.1% is achieved in the presence of slow movement and multiple persons at the relative distance of 100 cm.

Face Recognition under Varying Illumination Using Green’s Functionbased Bidimensional Empirical Mode Decomposition and Gradientfaces

A novel face recognition approach under varying illumination condition based on Green’s function in tension-based bid imensional empirical mode decomposition (GiT-BEMD) and gradient faces (GBEMDGF) is present. Firstly, face image was illumination normalization by discrete cosine transform that an appropriate number of DCT coefficients are truncated in logarithm domain. And then, two intrinsic mode functions (IMFs) that relevant of intrinsic physical significances of face images are produced by Gi T-BEMD. At the same time, gradient faces is used to improve the high frequency component of face images and to extract illumination insensitive facial feature. The facial feature of discriminately are fused using IMFs and illumination insensitive feature. Secondly, the principal component analysis is adopted to reduce the dimension of face image. The nearest neighbourhood classifier based on cosine distance is implemented for face classification. Experimental results on Yale B database and CUM PIE face database demonstrate that the present technique is robust to varying lighting resource.

Illumination invariant optical flow using neighborhood descriptors

Total variational (TV) methods using l1-norm are efficient approaches for optical flow determination. This contribution presents a multi-resolution TV-l1 approach using a data-term based on neighborhood descriptors and a weighted non-local regularizer. The proposed algorithm is robust to illumination changes. The benchmarking of the proposed algorithm is done with three reference databases (Middlebury, KITTI and MPI Sintel). On these databases, the proposed approach exhibits an optimal compromise between robustness, accuracy and computation speed. Numerous tests performed both on complicated data of the reference databases and on challenging endoscopic images acquired under three different modalities demonstrate the robustness and accuracy of the method against the presence of large or small displacements, weak texture information, varying illumination conditions and modality changes.

TRAFFIC SIGN DETECTION AND RECOGNITION: REVIEW AND ANALYSIS

Over the years, traffic sign detection and recognition systems gives extra value to driver assistance when driving, leading to more user-friendly driving experience and much improved safety for passengers. As part of Advanced Driving Systems (ADAS) one can be benefitted by using this system especially with driving incapacities by alerting and aid them about the existence of traffic signs to minimized unwanted circumstances during driving such as fatigue, poor sight and adverse weather conditions. Though a various number of traffic sign detection systems have been revised in literature; the need of design with a robust algorithm still remains open for further research. This paper purposes to design a system capable of performing traffic sign detection while considering variations of challenges such as color illumination, computational difficulty and functional constraints existed. Traffic sign detection is divided into three main parts namely; Pre-processing, Color segmentation and Thresholding. The color segmentation method is vital as it presents a detailed investigation of vision based color spaces in this case RGB, HSV and CMYK considering varying illumination conditions under different environments. This paper further highlights possible improvements to the proposed approaches for traffic sign detection.

Face Recognition in Complex Background: Developmental Network and Synapse Maintenance

As an important sub-field in computer vision and pattern recognition, face recognition has important theoretical and practical application. It is a very complicated problem which is often affected by variations in illumination condition, head pose, facial emotions, glasses, beards, and so on. 108 face images of 27 learned objects (appearances) from complex backgrounds in BioID face database are efficiently recognized with the developmental network (DN) - a biologically inspired framework with the emergent representation. But the DN has no adaptive receptive field for an object with a curved contour. Leaked-in background pixels can lead to problems when different objects look similar. This paper introduces another biologically inspired mechanism - synapse maintenance to achieve the object recognition. Synapse maintenance can automatically decide which synapse should be removed, kept or partial removed, thus it can weaken the complex background, strengthen the face features, reduce the bad influence of the complex background on the face recognition. Experimental results show that DN with the synapse maintenance can effectively recognize faces with complicated backgrounds and the recognition rate is over 95%.

A low cost outdoor testing facility for detailed photovoltaic device performance characterization

AbstractLong term outdoor performance monitoring is the ultimate method to capture the long term performance characteristics of PV devices. As time consuming as this process might be, it is the only way in which the local environmental condition versus PV device characteristics may be captured when making investment decisions. This data needs to be captured at least once through the seasonal cycle. It needs to be correlated to the environmental conditions, this monitoring climatic conditions is a requirement. Solar irradiation conditions must be monitored over a range of inclination angles – direct, diffuse and reflected components must be ideally captured separately. The panels must be held at maximum power generating point through the varying illumination conditions. Periodic IV traces must be captured, in addition to the power output to understand mechanisms of degradation, if present. Furthermore, the IV curves must be summarized in terms of relevant performance metrics. This work describes an outdoor testing facility that addresses all the aforementioned requirements. Reliable equipment selections coupled with adequate modelling strategies allow for the generation of a robust and accessible database of performance data. (© 2015 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)

Automatic moving object segmentation methods under varying illumination conditions for video data: comparative study, and an improved method

In recent past, many moving object segmentation methods under varying lighting changes have been proposed in literature and each of them has their own benefits and limitations. The various methods available in literature for moving object segmentation may be broadly classified into four categories i.e., moving object segmentation methods based on (i) motion information (ii) motion and spatial information (iii) learning (iv) and change detection. The objective of this paper is two-fold i.e., firstly, this paper presents a comprehensive comparative study of various classical as well as state-of-the art methods for moving object segmentation under varying illumination conditions under each of the above mentioned four categories and secondly this paper presents an improved approximation filter based method in complex wavelet domain and its comparison with other methods under four categories mentioned as above. The proposed approach consist of seven steps applied on given video frames which include: wavelet decomposition of frames using Daubechies complex wavelet transform; use of improved approximate median filter on detail co-efficient (LH, HL, HH); use of background modeling on approximate co-efficient (LL sub-band); soft thresholding for noise removal; strong edge detection; inverse wavelet transformation for reconstruction; and finally using closing morphology operator. The qualitative and quantitative comparative study of the various methods under four categories as well as the proposed method is presented for six different datasets. The merits, demerits, and efficacy of each of the methods under consideration have been examined. The extensive experimental comparative analysis on six different challenging benchmark data sets demonstrate that proposed method is performing better to other state-of-the-art moving object segmentation methods and is well capable of dealing with various limitations of existing methods.

Saliency guided naturalness enhancement in color images

Recent technological advances have enabled smartphones, tablet computers and wearable devices to acquire digital images conveniently. Thus, taking pictures, enjoying them, and using them in a variety of applications (e.g., social networking) have become an important part of technology based lifestyle for many people. As a result, efficient algorithms for image enhancement play an increasing role for improving image quality for both human visualization and computer image analysis. In this paper, we learn from the human visual system (HVS) which has a powerful biological mechanism to optimize scene perception in complex and varying illumination conditions. Using this knowledge, we combine the concept of popular Retinex and the histogram equalization (HE) and propose an efficient image naturalness enhancement algorithm for both non-uniform and low light images. Our algorithm emphasizes the perceptual contrast while reducing halo artifacts, detail-clipping effects, and over-enhancements. An input lightness is computed based on the white-patch assumption. A Retinex based lightness correction is utilized to preserve and improve the contrast for bright regions. The halo artifacts are suppressed based on Bilateral Filtering (BF). The detail-clipping effect is reduced by imposing certain constraints. For dark regions in input images, a perceptual contrast enhancement method with dynamic range adjustment is adopted which enhances perceptual contrast and adjusts dynamic range without over-enhancements. Finally, a weighting combination guided by a simplified saliency map is employed to blend results of enhancements. Our method produces more natural image appearance and better perceptual contrast in comparison with a number of state-of-the-art algorithms. Besides, the proposed algorithm is relatively simple and efficient in computation, suitable for implementation in mobile and wearable devices to support a variety of practical applications.

Saliency Map Generation by the Convolutional Neural Network for Real-Time Traffic Light Detection Using Template Matching

A critical issue in autonomous vehicle navigation and advanced driver assistance systems (ADAS) is the accurate real-time detection of traffic lights. Typically, vision-based sensors are used to detect the traffic light. However, the detection of traffic lights using computer vision, image processing, and learning algorithms is not trivial. The challenges include appearance variations, illumination variations, and reduced appearance information in low illumination conditions. To address these challenges, we present a visual camera-based real-time traffic light detection algorithm, where we identify the spatially constrained region-of-interest in the image containing the traffic light. Given, the identified region-of-interest, we achieve high traffic light detection accuracy with few false positives, even in adverse environments. To perform robust traffic light detection in varying conditions with few false positives, the proposed algorithm consists of two steps, an offline saliency map generation and a real-time traffic light detection. In the offline step, a convolutional neural network, i.e., a deep learning framework, detects and recognizes the traffic lights in the image using region-of-interest information provided by an onboard GPS sensor. The detected traffic light information is then used to generate the saliency maps with a modified multidimensional density-based spatial clustering of applications with noise (M-DBSCAN) algorithm. The generated saliency maps are indexed using the vehicle GPS information. In the real-time step, traffic lights are detected by retrieving relevant saliency maps and performing template matching by using colour information. The proposed algorithm is validated with the datasets acquired in varying conditions and different countries, e.g., USA, Japan, and France. The experimental results report a high detection accuracy with negligible false positives under varied illumination conditions. More importantly, an average computational time of 10 ms/frame is achieved. A detailed parameter analysis is conducted and the observations are summarized and reported in this paper.

A Machine Vision System for Quantification of Citrus Fruit Dropped on the Ground under the Canopy

Abstract. The overall goal of this study was to develop a machine vision system to quantify dropped citrus fruits on the ground. Specific objectives were: (1) to build a machine vision system suitable for citrus grove field conditions, (2) to develop an image enhancement algorithm for varying illumination conditions, and (3) to develop an image processing algorithm to estimate citrus fruit drop count and mass. The image processing algorithm consisted of (1) illumination enhancement using a retinex algorithm, (2) classification, (3) segmentation using a watershed algorithm with h-minima transform, and (4) ellipse fitting for mass estimation. Performances of the algorithms were evaluated in terms of correct identification and false positive errors. The average correct identification rate was 88.1%, 83.6%, and 82.9% for logistic regression, k-nearest neighbor (kNN), and Bayesian classifiers, respectively. False positive errors were 13.7%, 40.9%, and 17.9% for logistic regression, kNN, Bayesian classifiers, respectively. The results demonstrate the system‘s ability to quantify dropped fruits with specific geo-referenced location information. Spatially varied fruit drop maps plotted from the results can assist growers in finding problematic areas in their citrus groves more efficiently while reducing inspection and treatment costs. Such maps can also facilitate treatment of citrus Huanglongbing (HLB) disease in combination with HLB intensity data, psyllid counts, fertilization programs, and other block-specific management practices.

Estimating Snow Cover From Publicly Available Images

In this paper we study the problem of estimating snow cover in mountainous regions, that is, the spatial extent of the earth surface covered by snow. We argue that publicly available visual content, in the form of user generated photographs and image feeds from outdoor webcams, can both be leveraged as additional measurement sources, complementing existing ground, satellite and airborne sensor data. To this end, we describe two content acquisition and processing pipelines that are tailored to such sources, addressing the specific challenges posed by each of them, e.g., identifying the mountain peaks, filtering out images taken in bad weather conditions, handling varying illumination conditions. The final outcome is summarized in a snow cover index, which indicates for a specific mountain and day of the year, the fraction of visible area covered by snow, possibly at different elevations. We created a manually labelled dataset to assess the accuracy of the image snow covered area estimation, achieving 90.0% precision at 91.1% recall. In addition, we show that seasonal trends related to air temperature are captured by the snow cover index.

Local Feature Integration Method Using Phase Congruency for Palm Print Authentication

Palm print authentication is a biometric technology to identify a person's identity. In this paper, phase congruency method is used to extract features from palm print ROI images. The phase congruency is an efficient method to extract features at varying illumination condition and the image is invariant to contrast. By applying this method, local phase congruency (LPC), local orientation (LO) and local phase (LP) are extracted individually and fused using score level fusion. To reduce false acceptance rate (FAR), Min-Max threshold range is employed and the proposed method is tested on PolyU database of 7480 images from 374 individuals, with 20 image samples per individual. The proposed system achieves genuine acceptance rate (GAR) of 100% and FAR of 0.65%.

Multitarget Tracking in Nonoverlapping Cameras Using a Reference Set

Tracking multiple targets in nonoverlapping cameras are challenging since the observations of the same targets are often separated by time and space. There might be significant appearance change of a target across camera views caused by variations in illumination conditions, poses, and camera imaging characteristics. Consequently, the same target may appear very different in two cameras. Therefore, associating tracks in different camera views directly based on their appearance similarity is difficult and prone to error. In most previous methods, the appearance similarity is computed either using color histograms or based on pretrained brightness transfer function that maps color between cameras. In this paper, a novel reference set based appearance model is proposed to improve multitarget tracking in a network of nonoverlapping cameras. Contrary to previous work, a reference set is constructed for a pair of cameras, containing subjects appearing in both camera views. For track association, instead of directly comparing the appearance of two targets in different camera views, they are compared indirectly via the reference set. Besides global color histograms, texture and shape features are extracted at different locations of a target, and AdaBoost is used to learn the discriminative power of each feature. The effectiveness of the proposed method over the state of the art on two challenging real-world multicamera video data sets is demonstrated by thorough experiments.

Online Space-Variant Background Modeling With Sparse Coding

In this paper, we propose a sparse coding approach to background modeling. The obtained model is based on dictionaries which we learn and keep up to date as new data are provided by a video camera. We observe that, without dynamic events, video frames may be seen as noisy data belonging to the background. Over time, such background is subject to local and global changes due to variable illumination conditions, camera jitter, stable scene changes, and intermittent motion of background objects. To capture the locality of some changes, we propose a space-variant analysis where we learn a dictionary of atoms for each image patch, the size of which depends on the background variability. At run time, each patch is represented by a linear combination of the atoms learnt online. A change is detected when the atoms are not sufficient to provide an appropriate representation, and stable changes over time trigger an update of the current dictionary. Even if the overall procedure is carried out at a coarse level, a pixel-wise segmentation can be obtained by comparing the atoms with the patch corresponding to the dynamic event. Experiments on benchmarks indicate that the proposed method achieves very good performances on a variety of scenarios. An assessment on long video streams confirms our method incorporates periodical changes, as the ones caused by variations in natural illumination. The model, fully data driven, is suitable as a main component of a change detection system.

Flower colours in temperate forest and grassland habitats: a comparative study

Flower colours are important cues shaping the interaction between plants and pollinators. Plants flowering in distinct habitats like grasslands and forests present their flowers in very different light environments and against different backgrounds. Since most Angiosperms depend on or profit from pollination by animals, plants may exhibit flower colours that are most conspicuous when seen in the predominant illumination or against the predominant background colour of their environment. To compare flower colours of different habitats, we collected flower reflectance spectra of 239 herbs from forest and grassland sites in three German regions. We compared chromatic and achromatic components of flower colours from the honeybee’s point of view as well as in principal component analysis (PCA) to exclude the bias of particular visual systems. Our results show that flower colours do not differ between closed forest and open grassland habitats in any chromatic or achromatic aspect both from the bee’s perspective and without any model bias (PCA). Thus, although the colours of lights and backgrounds are different between both types of habitats, we find no evidence of an adaptation of flower colours to the visual system of bees. The finding that flower colours look similar for bees in different environments may be related to a mechanism called colour constancy which allows bees and other pollinator species to compensate for varying illumination conditions.

Validated Method for Repeatable Power Measurement of CIGS Modules Exhibiting Light-Induced Metastabilities

We report on the validation of a stabilization procedure designed to minimize variations in repeated power measurements at standard test conditions caused by transient light-induced metastabilities in copper indium gallium diselenide (CIGS) modules. Such metastable effects frustrate the repeatable and accurate measurement of a module's performance in the electrical state to which it stabilizes under normal operation outdoors. The procedure studied here is based on a light exposure followed by forward electrical bias as the module cools to the measurement temperature. The procedure was tested in a lab-to-lab intercomparison involving five different labs. Results show that the procedure is effective in yielding repeatable measurements and that the variations due to metastabilities are of roughly the same magnitude as those associated with variations in illumination conditions between different flash simulators. We also find that temperature-corrected measurements made immediately upon completion of the light exposure are less repeatable than those made after the module has cooled to 25°C under bias.

Face Recognition by PCA and Improved LBP Fusion Algorithm

This paper proposed a theoretically efficient approach for face recognition based on principal component analysis (PCA) and rotation invariant uniform local binary pattern texture features in order to weaken the effects of varying illumination conditions and facial expressions. Firstly, the rotation invariant uniform LBP operator was adopted to extract the local texture feature of the face images. Then PCA method was used to reduce the dimensionality of the extracted feature and get the eigenfaces. Finally, the nearest distance classification was used to distinguish each face. The method has been accessed on Yale and ATR-Jaffe face databases. Results demonstrate that the proposed method is superior to standard PCA and its recognition rate is higher than the traditional PCA. And the proposed algorithm has strong robustness against the illumination changes, pose, rotation and expressions.