Search

Most convolutional neural network (CNN)-based cloud detection methods are built upon the supervised learning framework that requires a large number of pixel-level labels. However, it is expensive and time-consuming to manually annotate pixelwise labels for massive remote sensing images. To reduce the labeling cost, we propose an unsupervised domain adaptation (UDA) approach to generalize the model trained on labeled images of source satellite to unlabeled images of the target satellite. To effectively address the domain shift problem on cross-satellite images, we develop a novel UDA method based on grouped features alignment (GFA) and entropy minimization (EM) to extract domain-invariant representations to improve the cloud detection accuracy of cross-satellite images. The proposed UDA method is evaluated on ``Landsat-8 → ZY-3'' and ``GF-1→ ZY-3'' domain adaptation tasks. Experimental results demonstrate the effectiveness of our method against existing state-of-the-art UDA approaches. The code of this paper has been made available online (https://github.com/nkszjx/grouped-features-alignment).

Automated segmentation of the optic disc (OD) and optic cup (OC) from different datasets plays an important role in the diagnosis of glaucoma and greatly saves human resources in both data annotation and image segmentation. However, the domain shift between different datasets suppresses the generalization ability of the segmentation network, especially damaging the performance of segmentation in the target domain, which is unlabeled. Therefore, using a transfer learning algorithm or domain adaptation method to enhance the migration ability of segmentation models has become an essential step and has attracted the attention of many researchers. In this paper, we propose an unsupervised domain adaptation network, called the Minimizing-entropy and Fourier Domain Adaptation network (MeFDA), to narrow the discrepancy between the source and target domains and prevent the degradation of segmentation performance. First, we perform adversarial optimization on the entropy maps of the predicted segmentation results to alleviate the domain shift. Then, direct entropy-minimization optimization is applied to the unlabeled target domain data to improve the credibility of the prediction segmentation maps. To enhance the prediction consistency of the target domain data, we augment the target domain dataset through the Fourier transform by replacing the low-frequency part in the target images with that of the source images. Then, a semantic consistency constraint is imposed on the raw images and augmented images of the target domain to improve the prediction consistency of the segmentation model, thereby further narrowing the discrepancy between the source and target domains. Experiments on several public retinal fundus image datasets prove the superiority of MeFDA compared with state-of-the-art methods, and the ablation study analyzes the importance of the different proposed components.

Localization microscopy offers resolutions down to a single nanometer, but currently requires additional dedicated hardware or fiducial markers to reduce resolution loss from drift of the sample. Drift estimation without fiducial markers is typically implemented using redundant cross correlation (RCC). We show that RCC has sub-optimal precision and bias, which leaves room for improvement. Here, we minimize a bound on the entropy of the obtained localizations to efficiently compute a precise drift estimate. Within practical compute-time constraints, simulations show a 5x improvement in drift estimation precision over the widely used RCC algorithm. The algorithm operates directly on fluorophore localizations and is tested on simulated and experimental datasets in 2D and 3D. An open source implementation is provided, implemented in Python and C++, and can utilize a GPU if available.

Pseudo-labeling is a simple and well known strategy in Semi-Supervised Learning with neural networks. The method is equivalent to entropy minimization as the overlap of class probability distribution can be reduced minimizing the entropy for unlabeled data. In this paper we review the relationship between the two methods and evaluate their performance on Fine-Grained Visual Classification datasets. We include also the recent released iNaturalist-Aves that is specifically designed for Semi-Supervised Learning. Experimental results show that although in some cases supervised learning may still have better performance than the semi-supervised methods, Semi Supervised Learning shows effective results. Specifically, we observed that entropy-minimization slightly outperforms a recent proposed method based on pseudo-labeling.

We present a novel approach for unsupervised road segmentation in adverse weather conditions such as rain or fog. This includes a new algorithm for source-free domain adaptation (SFDA) using self-supervised learning. Moreover, our approach uses several techniques to address various challenges in SFDA and improve performance, including online generation of pseudo-labels and self-attention as well as use of curriculum learning, entropy minimization and model distillation. We have evaluated the performance on $6$ datasets corresponding to real and synthetic adverse weather conditions. Our method outperforms all prior works on unsupervised road segmentation and SFDA by at least 10.26%, and improves the training time by 18-180x. Moreover, our self-supervised algorithm exhibits similar accuracy performance in terms of mIOU score as compared to prior supervised methods.

This paper presents a new entropy minimization criterion for both symbol timing and carrier frequency recovery for wireless receivers. Synchronization is achieved by minimizing the entropy estimated from the eye diagram and the constellation diagram. Key implementation details are addressed toward the realization of entropy-based synchronization algorithms. In addition, the performance is evaluated in controlled conditions. It is shown that, as an alternative to the maximum likelihood criterion, entropy minimization has great potential and offers certain advantages for synchronization in wireless communication, particularly for pulse shaping filters with small excess bandwidth, as well as in multipath fading channels.

Unsupervised domain adaptation algorithms seek to transfer knowledge from labeled source datasets in order to predict the labels for target datasets in the presence of domain-shift. In this paper we propose the Certain and Consistent Domain Adaptation (CCDA) model for unsupervised domain adaptation. The CCDA aligns the source and target domains using adversarial training and reduces the domain adaptation problem to a semi supervised learning (SSL) problem. We estimate the target labels using consistency regularization and entropy minimization on the domain-aligned target samples whose predictions are consistent across multiple stochastic perturbations. We evaluate the CCDA on benchmark datasets and demonstrate that it outperforms competitive baselines from domain adaptation literature.

Partial label learning deals with the problem where each training example is associated with a set of candidate labels, only one of which is assumed to be valid. To learn from such ambiguous labeling information, the critical point is to disambiguate the set of candidate labels, thereby targeting the ground-truth label. By utilizing the nature that only one of the candidate labels is correct, we employ the entropy minimization strategy to force the model making confident predictions of the training data. By doing this, the ground-truth labels are likely to make more contributions to the model training. Finally, comparative experiments on a number of real-world datasets are conducted, demonstrating the effectiveness of the proposed approach.

In this paper, we address the problem of retrospective color shading correction. An extension of the established gray-level shading correction algorithm based on signal envelope (SE) estimation to color images is developed using principal color components. Compared to the probably most general shading correction algorithm based on entropy minimization, SE estimation does not need any computationally expensive optimization and thus can be implemented more effciently. We tested our new shading correction scheme on artificial as well as real endoscopic images and observed promising results. Additionally, an indepth analysis of the stop criterion used in the SE estimation algorithm is provided leading to the conclusion that a fixed, user-defined threshold is generally not feasible. Thus, we present new ideas how to develop a non-parametric version of the SE estimation algorithm using entropy.

Fully convolutional neural networks (FCN) are successfully used for the automated pixel-wise classification of aerial images and possibly additional data. However, they require many labelled training samples to perform well. One approach addressing this issue is semi-supervised domain adaptation (SSDA). Here, labelled training samples from a source domain and unlabelled samples from a target domain are used jointly to obtain a target domain classifier, without requiring any labelled samples from the target domain. In this paper, a two-step approach for SSDA is proposed. The first step corresponds to a supervised training on the source domain, making use of strong data augmentation to increase the initial performance on the target domain. Secondly, the model is adapted by entropy minimization using a novel weighting strategy. The approach is evaluated on the basis of five domains, corresponding to five cities. Several training variants and adaptation scenarios are tested, indicating that proper data augmentation can already improve the initial target domain performance significantly resulting in an average overall accuracy of 77.5%. The weighted entropy minimization improves the overall accuracy on the target domains in 19 out of 20 scenarios on average by 1.8%. In all experiments a novel FCN architecture is used that yields results comparable to those of the best-performing models on the ISPRS labelling challenge while having an order of magnitude fewer parameters than commonly used FCNs.

Unsupervised domain adaptation has achieved great progress in the past few years. Nevertheless, most existing methods work in the so-called closed-set scenario, assuming that the classes depicted by the target samples are exactly the same as those of the source domain. In this paper, we tackle the more challenging scenario of open set domain adaptation with a novel end-to-end training approach, where the samples of unknown class can be present in the target domain. Our method employs entropy minimization for performing unsupervised domain adaptation, where unknown samples are aggressively used in training by forcing the classifier to output the probability of 0.5 on the unknown class. Experimental evidence demonstrates that our approach significantly outperforms the state-of-the-art in open set domain adaptation.

Semantic segmentation is a key problem for many computer vision tasks. While approaches based on convolutional neural networks constantly break new records on different benchmarks, generalizing well to diverse testing environments remains a major challenge. In numerous real world applications, there is indeed a large gap between data distributions in train and test domains, which results in severe performance loss at run-time. In this work, we address the task of unsupervised domain adaptation in semantic segmentation with losses based on the entropy of the pixel-wise predictions. To this end, we propose two novel, complementary methods using (i) entropy loss and (ii) adversarial loss respectively. We demonstrate state-of-the-art performance in semantic segmentation on two challenging "synthetic-2-real" set-ups and show that the approach can also be used for detection., Accepted in CVPR'19. Code is available at https://github.com/valeoai/ADVENT

Optical scanning holography (OSH) is a powerful and effective method for capturing the complex hologram of a three-dimensional (3-D) scene. Such captured complex hologram is called optical scanned hologram. However, reconstructing a focused image from an optical scanned hologram is a difficult issue, as OSH technique can be applied to acquire holograms of wide-view and complicated object scenes. Solutions developed to date are mostly computationally intensive, and in so far only reconstruction of simple object scenes have been demonstrated. In this paper we report a low complexity method for reconstructing a focused image from an optical scanned hologram that is representing a 3-D object scene. Briefly, a complex hologram is back-propagated onto regular spaced images along the axial direction, and from which a crude, blocky depth map of the object scene is computed according to non-overlapping block partitioned entropy minimization. Subsequently, the depth map is low-pass filtered to decrease the blocky distribution, and employed to reconstruct a single focused image of the object scene for extended depth of field. The method proposed here can be applied to any complex holograms such as those obtained from standard phase-shifting holography. Published version

To present focused ISAR imaging results in the homogenous range and cross-range domain, an integrated scheme is proposed to estimate both the targets equivalent rotational velocity (RV) and rotational center (RC). The RV estimation is improved by radial projection combined with keystone processing, and then the RC is estimated through image entropy minimization. Finally, delicate imaging results may be obtained for wide-angle scenarios. Experiment results are provided to demonstrate the effectiveness of the proposed method.

Discretization of numerical data is one of the most influential data preprocessing tasks in knowledge discovery and data mining. The purpose of attribute discretization is to find concise data representations as categories which are adequate for the learning task retaining as much information in the original continuous attribute as possible. In this article, we present an updated overview of discretization techniques in conjunction with a complete taxonomy of the leading discretizers. Despite the great impact of discretization as data preprocessing technique, few elementary approaches have been developed in the literature for Big Data. The purpose of this article is twofold: a comprehensive taxonomy of discretization techniques to help the practitioners in the use of the algorithms is presented; the article aims is to demonstrate that standard discretization methods can be parallelized in Big Data platforms such as Apache Spark, boosting both performance and accuracy. We thus propose a distributed implementation of one of the most well-known discretizers based on Information Theory, obtaining better results than the one produced by: the entropy minimization discretizer proposed by Fayyad and Irani. Our scheme goes beyond a simple parallelization and it is intended to be the first to face the Big Data challenge. WIREs Data Mining Knowl Discov 2016, 6:5-21. doi: 10.1002/widm.1173

We propose an information-theoretic criterion, entropy estimate, for the joint alignment of a group of shape observations drawn from an unknown shape distribution. Employing a nonparametric density estimation technique with implicit shape representation, we minimize the entropy estimate with respect to the pose parameters of similarity transformations based on gradient descent optimization for which we provide implementation details. We demonstrate the capacity of our approach in numerous experiments with an application of building a shape prior to prostate MR image segmentation.

Autofocusing technology is an essential step of Inverse Synthetic Aperture Radar (ISAR) imaging, whose performance has great influence on the quality of the radar image. As far as the existed autofocusing methods are concerned, the methods based on minimum entropy criterion are robust and have been widely applied in both Synthetic Aperture Radar (SAR) and ISAR imaging. However, the Minimum Entropy based Autofocusing (MEA) methods usually suffer from heavy computation burden because of the complex formula of image entropy and the optimal search of the phase error. In this paper, a novel fast MEA method based on the Newton method is proposed. Moreover, the Simplified Newton (SN) method and the Modified Newton (MN) method are also introduced to the implement of MEA, and provide high computation efficiency. Mathematical analysis and experimental results based on real measured data of two airplanes have validated the high computation efficiency of the proposed methods. Compared to the widely applied MEA method for ISAR imaging, the proposed methods can improve the computation efficiency by 10 to 20 times.

We propose a novel consistency based regularization method for semi-supervised image classification, called feature distribution matching (FDM), which is designed to induce smoothness of feature space by reducing sliced Wasserstein distance between feature distributions of labeled and unlabeled set. Unlike previous perturbation based methods, FDM does not require extra computational cost except one regularization loss. Our result shows that FDM combined with entropy minimization improves classification accuracy compared to supervised-only baseline and some previous methods. We also analyze our method by visualizing feature embeddings which shows that FDM lead smooth data manifold on feature space.

Open-path Fourier transform infrared (OP/FT-IR) spectra were arranged into data matrices and then analyzed with target factor analysis (TFA) and band-target entropy minimization (BTEM), respectively, aiming to reconstruct the spectra of interested molecules. Five sessions of continuous OP/FT-IR monitoring were carried out around farms. For each session, the spectra were arranged row-wisely in the order of measuring time, which yielded five matrices for data analysis. The analyses results showed that both TFA and BTEM could reconstruct spectral features of target molecules from the spectra data matrix, but performance of the two methods differed slightly. TFA can retrieve spectral features of target molecules in the presence of interferences, and the reconstructed spectrum is similar to corresponding reference. BTEM can implement such spectral retrieval without the reference spectrum. This paper presents not only the application of BTEM method to qualitative analyses of OP/FT-IR spectra, but also a thorough comparison between TFA and BTEM, which is helpful to qualitative analysis of complex multi-component systems.

Breast cancer is especially a concern in woman and it is one of the most leading causes of death among woman in the world. The adoption of classical statistical modeling techniques has standardized computer vision in last few years. For better recognition information, most recognition algorithms depend on careful positioning of an object into a canonical pose, so that we can have meaningful information examined. Currently, a new statistical-theoretic approach is presented for finding the alignment (using entropy minimization with brightness transformation) of mammographic images of differing patients in this paper. Registration is achieved by adjusting the relative position and orientation until the sum of the pixel-stack entropies is minimized. Segmentation analysis of digital or digitized mammographic images was firstly carried out on mammographic images of left and right breast images for 161 patients, obtained from Mammographic Image Analysis Society (MIAS) database. Given a set of unaligned exemplars of a class, such as breast, we applied an automatically built alignment mechanism, no additional labeling of parts or poses in the dataset. Using this alignment mechanism, new members of the class, such as an unseen mammographic image from a new patient, can be precisely aligned for the recognition process. Our alignment method can improve the performance on a breast abnormality recognition task, both over unaligned images and over images aligned with the proposed alignment algorithm.

In situations where available information or evidence is incomplete or uncertain, probabilistic two-way decisions/classifications with a single threshold on probabilities for making either an acceptance or a rejection decision may be inappropriate. With the introduction of a third non-commitment option, probabilistic three-way decisions use a pair of thresholds and provide an effective and practical decision-making strategy. This paper presents a multifaceted analysis of probabilistic three-way decisions. By identifying an inadequacy of two-way decisions with respect to controlling the levels of various decision errors, we examine the motivations and advantages of three-way decisions. We present a general framework for computing the required thresholds of a three-way decision model as an optimization problem. We investigate two special cases, one is a decision-theoretic rough set model and the other is an information-theoretic rough set model. Finally, we propose a heuristic algorithm for finding the required thresholds.

In this paper, we propose a robust probability based sparse method to solve single sample face recognition, which harvests the advantages of both local and global representation. Different from previous sparse representation methods that generate sparse coefficients by 𝑙1, we produce sparse class probability distribution by proposing a multi-phase sparse probability (MSP) framework. To create class probability distribution, we divide each face image into many local blocks and vote based on the classification results of all blocks. For classifying each block, we propose local similarity assumption that makes many conventional methods feasible to SSPP problem. Moreover, we also propose a heuristic multiphase class selection scheme to solve the entropy minimization problem, which finally provides a higher classification confidence from the global perspective. Experimental results on three popular databases show that our approach not only generalizes well to SSPP problem but also has strong robustness to expression, illumination, occlusion and time variation.

We investigated a novel fast non-uniform illumination correction method for bi-level images. The proposed method divides a bi-level image into sub-images and roughly estimates block-wise illumination by low pass filtered maximum values of sub-images. After that, we apply bilinear interpolation using the block-wise illumination to estimate non-uniform illumination, and compensate for the effect of non-uniform illumination using the estimated illumination. Since the proposed method is not based on computation intensive iterative optimization, the proposed method can be used effectively for applications that require fast correction of non-uniform illumination. In simulations, the proposed method showed more than 20 times faster speed than existing entropy minimization method. Moreover, in simulations and experiments, the restored images by the proposed method were more close to true images than images restored by conventional method.

In recent decades, the development and application of expert systems has been under intensive study (e.g., see [1---3]). The construction of an optimal algorithm for guessing an object by sequentially asking questions about its properties in the case when the answers may be incorrect is considered. This algorithm can be used in expert systems. Several optimization criteria are considered. It is concluded that the most practical criterion is the entropy minimization. The one-step entropy minimization is an analog of a deterministic greedy algorithm. The question asked by the expert system that minimizes the entropy can be chosen at each step of the procedure for any search depth depending on available computational resources.

Most of the existing stereo matching algorithms will assume a similar corresponding color values between stereo images. In the real scenario, these color values are effected by several radiometric factors such as illumination direction, illumination color, camera parameters, etc, which results in different color values between the corresponding points. Hence, applying the stereo algorithm directly on the raw color values is not appropriate for the real-time environment. This paper proposes an entropy minimization based log chromaticity projection for stereo image, thereby extracting the invariant image, which is independent of illumination and color. The developed invariant image is the perfect measure for finding the similarity between the corresponding points. Normalized cross correlation based similarity measure is applied on the generated invariant image and the obtained disparity outperforms some of the local and global stereo algorithms.

Many scholars have attempted to use an encryption method to resolve the problem of data leakage in data outsourcing storage. However, encryption methods reduce data availability and are inefficient. Vertical fragmentation perfectly solves this problem. It was first used to improve the access performance of the relational database, and nowadays some researchers employ it for privacy protection. However, there are some problems that remain to be solved with the vertical fragmentation method for privacy protection in the relational database. First, current vertical fragmentation methods for privacy protection require the user to manually define privacy constraints, which is difficult to achieve in practice. Second, there are many vertical fragmentation solutions that can meet privacy constraints, however, there are currently no quantitative evaluation criteria evaluating how effectively solutions can protect privacy more effectively. In this article, we introduce the concept of information entropy to quantify privacy in vertical fragmentation, so we can automatically discover privacy constraints. Based on this, we propose a privacy protection model with a minimum entropy fragmentation algorithm to achieve minimal privacy disclosure of vertical fragmentation. Experimental results show that our method is suitable for privacy protection with a lower overhead.

The present contribution reviews the operando spectroscopy programme in the area of homogeneous catalysis that exists at the Institute of Chemical and Engineering Sciences (ICES). This review first considers the importance of experimental design and then addresses a number of signal processing issues. Band-target entropy minimization (BTEM), an algorithm designed for pure component spectral estimation, is the most important signal processing tool discussed. A homogeneous catalytic example using BTEM analysis, as well as other quantitative and kinetic analyses is shown. In addition to spectroscopic measurements, other in situ sensor measurements and their analyses are also discussed. The determination of individual physico-chemical properties of the constituents present allows insight into solute–solute interactions, etc. This review is based on a plenary lecture presented at Operando III.

We propose a tracking method based on minimization of multi-detector trajectory entropy. Multiple detectors are constructed online to track objects synchronously by mutual information-based feature selection. Each detector obtains its own tracking trajectory through continuous detection. The tracking results of all detectors are combined based on the minimization of trajectory entropy, and the optimal detector is selected from it to determine the target trajectory and achieve the target tracking task. Experimental results show that our tracker can handle more complex tracking environments and outperform many other state-of-the-art methods in terms of both success rate and precision.

Test selection in diagnosis is a procedure suggesting tests to be executed when trying to answer the query "What is the diagnosis for this problem?". However, other queries, such as "Is h the diagnosis for this problem?", are relevant as they can involve faster test selection algorithms and they can result in a lower test execution cost. Usually, a one step lookahead entropy minimization strategy is adopted to implement the test selection procedure. However, we show that this strategy can be arbitrarily bad and therefore, it is important to consider several strategies to solve a query. Each strategy taking a different position in the tradeoff computation time vs test execution cost. In this paper, we consider a query-based approach where test selection is justified and driven by a user's specific query. We also study different strategies, optimal and approximate, for test selection. Finally, we illustrate how the operating system discovery (OSD) problem can be solved using a diagnosis framework and how it benefits from a query-based approach.

Compressed sensing (CS) provides a new method to encode depth videos, which utilizes the sparsity of depth maps to improve the coding efficiency. In this paper, we design a novel CS-based depth video codec. The codec adaptively decomposes blocks from 64 χ 64 to 8 χ 8 sub-blocks via wavelet transforming. The decomposition divides smooth regions and complex boundaries by frequencies, and minimizes the local entropy of sub-blocks. Moreover, a measurement rate allocation algorithm is also proposed, which utilizes the rate-distortion optimization (RDO) to allocate the measurement rate for each block. The experimental results demonstrate that, compared with H.264/AVC and H.265/HEVC, the proposed codec improves the quality of virtual views by 1–2 dB and 0.2–0.5 dB PSNR respectively. Meanwhile, the codec reduces coding complexity greatly.

In this paper, we propose a new approach to build the background image which is the first step for the detection and identification of moving objects in a video. This approach is of global type and proceeds to split the video images into regions. The criterion of entropy minimization over time is adopted to ensure if the considered region would participate in the composition of the background image. Many experiments demonstrate the utility and performance of this approach.

International audience; Air traffic management (ATM) aims at providing companies with a safe and ideally optimal aircraft trajectory planning. Air traffic controllers act on flight paths in such a way that no pair of aircraft come closer than the regulatory separation norm. With the increase of traffic, it is expected that the system will reach its limits in a near future: a paradigm change in ATM is planned with the introduction of trajectory based operations. This paper investigate a mean of producing realistic air routes from the output of an automated trajectory design tool. For that purpose, an entropy associated with a system of curves is defined and a mean of iteratively minimizing it is presented. The network produced is suitable for use in a semi-automated ATM system with human in the loop.

Statistical relationships between human vision and natural scene statistics have been addressed. Contrast is fundamental image property. We proposed that infant visual exploration and learning development could be analyzed using contrast entropy minimization. Ten infants viewed 16 naturalistic images. Fixations of the infants were recorded. Infant gaze patterns are compared between individual infants. Infants are compared to the fixations predicted by the contrast entropy minimization. The results show that infant tends to have different gaze patterns, and infant entropy drops gradually as fixations are made.

Mobile light detection and ranging (LiDAR) sensors used in car navigation and robotics, such as the Velodyne’s VLP-16 and HDL-32E, allow for sensing the surroundings of the platform with high temporal resolution to detect obstacles, tracking objects and support path planning. This study investigates the feasibility of using LiDAR sensors for tracking taxiing or landing aircraft close to the ground to improve airport safety. A prototype system was developed and installed at an airfield to capture point clouds to monitor aircraft operations. One of the challenges of accurate object tracking using the Velodyne sensors is the relatively small vertical field of view (30°, 41.3°) and angular resolution (1.33°, 2°), resulting in a small number of points of the tracked object. The point density decreases with the object–sensor distance, and is already sparse at a moderate range of 30–40 m. The paper introduces our model-based tracking algorithms, including volume minimization and cube trajectories, to address the optimal estimation of object motion and tracking based on sparse point clouds. Using a network of sensors, multiple tests were conducted at an airport to assess the performance of the demonstration system and the algorithms developed. The investigation was focused on monitoring small aircraft moving on runways and taxiways, and the results indicate less than 0.7 m/s and 17 cm velocity and positioning accuracy achieved, respectively. Overall, based on our findings, this technology is promising not only for aircraft monitoring but for airport applications.

Purpose: Improving the shape statistics of medical image objects by generating correspondence of interior skeletal points. Data: Synthetic objects and real world lateral ventricles segmented from MR images. Method(s): Each object’s interior is modeled by a skeletal representation called the s-rep, which is a quadrilaterally sampled, folded 2-sided skeletal sheet with spoke vectors proceeding from the sheet to the boundary. The skeleton is divided into three parts: up-side, down-side and fold-curve. The spokes on each part are treated separately and, using spoke interpolation, are shifted along their skeletal parts in each training sample so as to tighten the probability distribution on those spokes’ geometric properties while sampling the object interior regularly. As with the surface-based correspondence method of Cates et al., entropy is used to measure both the probability distribution tightness and sampling regularity. The spokes’ geometric properties are skeletal position, spoke length and spoke direction. The properties used to measure the regularity are the volumetric subregions bounded by the spokes, their quadrilateral sub-area and edge lengths on the skeletal surface and on the boundary. Results: Evaluation on synthetic and real world lateral ventricles demonstrated improvement in the performance of statistics using the resulting probability distributions, as compared to methods based on boundary models. The evaluation measures used were generalization, specificity, and compactness. Conclusions: S-rep models with the proposed improved correspondence provide significantly enhanced statistics as compared to standard boundary models.

We present a novel group-wise registration method for cortical correspondence for local cortical thickness analysis in human and non-human primate neuroimaging studies. The proposed method is based on our earlier template based registration that estimates a continuous, smooth deformation field via sulcal curve-constrained registration employing spherical harmonic decomposition of the deformation field. This pairwise registration though results in a well-known template selection bias, which we aim to overcome here via a group-wise approach. We propose the use of an unbiased ensemble entropy minimization following the use of the pairwise registration as an initialization. An individual deformation field is then iteratively updated onto the unbiased average. For the optimization, we use metrics specific for cortical correspondence though all of these are straightforwardly extendable to the generic setting: The first focused on optimizing the correspondence of automatically extracted sulcal landmarks and the second on that of sulcal depth property maps. We further propose a robust entropy metric and a hierarchical optimization by employing spherical harmonic basis orthogonality. We also provide the detailed methodological description of both our earlier work and the proposed method with a set of experiments on a population of human and non-human primate subjects. In the experiment, we have shown that our method achieves superior results on consistency through quantitative and visual comparisons as compared to the existing methods.

In view of the limitation of Gaussian feature of the data in traditional FCA model, a novel approach based on controlling the shape of probability density function (PDF) of the recovering error and entropy minimization to determine the number of principal components is proposed in this paper. The unique variable of the PCA model is the number of principal components (PCs). The PDF of the recovering error is made to track the given narrow Gaussian-distribution PDF and the PC number is determined to make the tracking error be minimized. If the target PDF is unknown, the entropy of the recovering error is minimized to determine the number of principal components. For this approach, the limitation of the recovering error obeying Gaussian distribution required by mean squared error (MSE) criterion is eliminated. As the result, the application range of the PCA model is extended. Computer simulations illustrate the effectiveness of the proposed approach.

The eyes are important facial landmarks, both for image normalization due to their relatively constant interocular distance, and for post processing due to the anchoring on model-based schemes. This paper introduces a novel approach for the eye detection task using optimal wavelet packets for eye representation and Radial Basis Functions (RBFs) for subsequent classification ("labeling") of facial areas as eye versus non-eye regions. Entropy minimization is the driving force behind the derivation of optimal wavelet packets. It decreases the degree of data dispersion and it thus facilitates clustering ("prototyping") and capturing the most significant characteristics of the underlying (eye regions) data. Entropy minimization is thus functionally compatible with the first operational stage of the RBF classifier, that of clustering, and this explains the improved RBF performance on eye detection. Our experiments on the eye detection task prove the merit of this approach as they show that eye images compressed using optimal wavelet packets lead to improved and robust performance of the RBF classifier compared to the case where original raw images are used by the RBF classifier.

We consider a slotted wireless network in an infrastructure setup with a base station (or an access point) and N users. The wireless channel gain between the base station and the users is assumed to be i.i.d. over users and slots, and the base station seeks to schedule the user with the highest channel gain in every slot (opportunistic scheduling). We assume that contention for opportunistic scheduling is resolved using a series of minislots and with feedback from the base station. In this setup, we formulate the contention resolution problem for opportunistic scheduling as identifying a random threshold (channel gain) that separates the best channel from the other samples. The average delay minimization for contention resolution is then related to entropy (of the random threshold) minimization, which is a concave minimization problem. We illustrate our formulation by studying a popular contention resolution strategy called the opportunistic splitting algorithm (OSA, [9]). OSA is a greedy algorithm that maximizes the probability of success in every minislot. We study the delay and entropy optimality of OSA for i.i.d. wireless channel. Finally, we discuss the applicability of the entropy minimization framework to identify optimal contention resolution strategies for general network scenarios.

Data from sensor logs in raw form is generally continuous valued. This data from multiple sensors in continuous stream becomes voluminous. For knowledge discovery like extraction of context, from these datasets, standard machine learning algorithms or their variations are used as classifier. Most classification schemes require the input data to be discretized. The focus of this paper is to study merits of some popular discretization methods when applied on noisy sensor logs. Representative methods from supervised and unsupervised discretization, like binning, clustering and entropy minimization are evaluated with context extraction. Interestingly, unlike common perception, for discretization of sensor data, supervised algorithms do not have a clear edge over unsupervised.

In this paper, we discuss a novel scheme for arriving at a sparse representation and recovery of a class of signals using information theoretic measures. Constituent components containing distinct features of any signal, belonging to a specific class, are separated and represented sparsely in an appropriate fixed basis. The morphological correlation between each of the constituent components and a subset of basis leads to sparse representation of the signal in that basis. The basis is selected using entropy minimization based method which is known to result in coefficient concentration. Simulation studies on speech signals show that in the presence of input noise, the proposed method outperforms conventional methods.

Wavelet filtering is combined with cyclostationary analysis for detection of gear tooth faults in a gearbox. The parameters of the wavelet filter are optimized by using the proposed entropy minimization rule. This method is shown to be effective in detecting gear faults when cyclostationary analysis by itself fails. @DOI: 10.1115/1.1760565#

There are two major issues of automatic entity relation extraction: human intervention and difficulty in labeling corpus. For these two problems, combined with the characteristics of the tourism domain, this paper adopts a weakly-supervised extraction method of entity relation based on entropy minimization. This method firstly extracts the characteristic words by the idea of scalar clustering with small-scale stratified marked instances, and constructs the initial classifier with maximum entropy machine learning algorithm. Then use the initial classifier of certain accuracy to classify the unlabeled instances, and add the instances of the minimum information entropy to the training corpus set to continually expand the scale of training data set. Finally, repeat the above iterative process until the performance of classifier is to be stabilized, and then a final extraction classifier of entity relation in specific domain will be constructed. Experiments performed on the corpus of tourism domain show that, not only can this method reduce the dependence of entity relation extraction on manual intervention, but it could effectively improve the performance of entity relation extraction, the F value of which is up to 63.69 %.

A rotation correction method of dorsal-hand vein (DHV) images is proposed. The idea of the proposed method is to normalize all the samples to the same virtual template defined with a criterion, so that we do not need to estimate the rotation angle between a sample and every template one by one in feature matching. The proposed method contains two main procedures, i.e., boundary detection and entropy minimization. Boundary detection is used to create a boundary map, and entropy minimization is adopted to find the best rotation angle with the boundary map. Simulations and real experiments are given to show that the proposed method can be completed in 65 milliseconds and can reduce equal error rate (EER) efficiently when the rotation angle in a database is big.

Fluorescence loss in photobleaching (FLIP) is a method to study compartment connectivity in living cells. A FLIP sequence is obtained by alternatively bleaching a spot in a cell and acquiring an image of the complete cell. Connectivity is estimated by comparing fluorescence signal attenuation in different cell parts. The measurements of the fluorescence attenuation are hampered by the low signal to noise ratio of the FLIP sequences, by sudden sample shifts and by sample drift. This paper describes a method that estimates the attenuation by modeling photobleaching as exponentially decaying signals. Sudden motion artifacts are minimized by registering the frames of a FLIP sequence to target frames based on the estimated model and by removing frames that contain deformations. Linear motion (sample drift) is reduced by minimizing the entropy of the estimated attenuation coefficients. Experiments on 16 in vivo FLIP sequences of muscle cells in Drosophila show that the proposed method results in fluorescence attenuations similar to the manually identified gold standard, but with standard deviations of approximately 50 times smaller. As a result of this higher precision, cell compartment edges and details such as cell nuclei become clearly discernible. The main value of this method is that it uses a model of the bleaching process to correct motion and that the model based fluorescence intensity and attenuation estimates can be interpreted easily. The proposed method is fully automatic, and runs in approximately one minute per sequence, making it suitable for unsupervised batch processing of large data series.

This paper proposes a novel framework for image classification with an entropy based image semantic cycle. Entropy minimization leads to an optimal image semantic cycle where images are connected in the semantic order. For classification, the training step is to find an optimal image semantic cycle in an image database. In the test step, the suitable position of an unknown image in this cycle is first found. Then, the class membership is determined through recognizing the nearest neighbors at this position. Experimental results demonstrate that the proposed framework achieves higher classification accuracy.

This paper presents an active vision approach to enhance mobile robot localization. A particle filter localization is extended with a module to find active vision decisions that are optimal based on the current localization and its uncertainty. Optimality is expressed as a criterion of entropy minimization. Further approximations are introduced to enable real-time computation. Both the usefulness of the presented approach in a RoboCup scenario and the performance and quality of the approximations are evaluated in different static and dynamic situations.

The goal in this paper is to detect and remove shadows of moving objects in colour image sequences. The method works by deriving the shadow edges of a 3-band colour image, integrating edge information temporally in the image sequence, and subsequently removing these shadows by a method paralleling reflectance recovery by entropy minimisation. The method incorporates background subtraction information and extensive edge-linking morphology to detect superior shadow edges. This integration has greatly increased the scope of applicability and brought about significant enhancements in the recovered shadow-free images and the time of processing. Shadows in the frames of image sequences are shown to be significantly attenuated. At present the implementation runs about 4 frames per second and can serve as a good utility in improving the performance of high level algorithms in scene monitoring. Refereed/Peer-reviewed

We speculate on how computational strategies suggested by Information theory could be useful for knowledge acquisition in sensory systems. The focus of our exploration is the idea that perception is a data reduction problem and as such sensory transformations should be predictable from the data reduction problem and as such sensory transformations should be predictable terms of variational principles involving the minimization of two types of entropy. The analysis suggests a scheme in which the two can be combined to produce a predictive design principle for sensory pathways.