Local density based background estimation

Background estimation and removal based on the joint use of range and color data produces superior results than can be achieved with either data source alone. This is increasingly relevant as inexpensive, real-time, passive range systems become more accessible through novel hardware and increased CPU processing speeds. Range is a powerful signal for segmentation which is largely independent of color and hence not effected by the classic color segmentation problems of shadows and objects with color similar to the background. However range alone is also not sufficient for the good segmentation: depth measurements are rarely available at all pixels in the scene, and foreground objects may be indistinguishable in depth when they are close to the background. Color segmentation is complementary in these cases. Surprisingly, little work has been done to date on joint range and color segmentation. We describe and demonstrate a background estimation method based on a multidimensional (range and color) clustering at each image pixel. Segmentation of the foreground in a given frame is performed via comparison with background statistics in range and normalized color. Important implementation issues such as treatment of shadows and low confidence measurements are discussed in detail.

A new background estimation and suppression algorithm was presented. In the algorithm, targets and observing noises were considered as mixed interferences of the image background. With this situation, image background was estimated adaptively and then background suppression was done in order to improve the signal-to-noise ratio (SNR) of targets. In this algorithm, firstly, a Zernike-facet model of image background was built up. Secondly, the total least squares (TLS) method was used to solve parameters of the model. Finally, background estimation and suppression were done using the model and its parameters. Simulations and several experiments demonstrating the effectiveness of this proposed algorithm were reported. And results show that this algorithm can be effective to estimate background in mixed noise environment and can preserve detail information of targets and improve SNR of targets. As a result, detecting probability and false probability will be improved in next process for automatic target detection and tracking.

Unstructured detectors such as KGLRT, ACE and AMF are widely applied for target detection in hyperspectral imagery (HSI). However, conventional global and local approaches construct background model without considering the contamination caused by anomalies and suspected targets. This paper proposes a local ACE algorithm based on the minimum covariance determinant (MCD) estimator. In the proposed algo-rithm, a spectral angle based clustering method is applied to the whitened hyperspectral data to form several disjoint clusters over the whole image. Then for each cluster, the robust estimations of its background statistics are obtained using the MCD estimator. Finally, the ACE detector is applied to each pixel utilizing the robust background statistics of the cluster. With experimental results on two different real datasets, the superiority of the proposed algorithm is demonstrated.

Edge detection and recognition is one of the most critical procedures in image processing and pattern recognition. Traditional Canny algorithm has some advantages, such as simple and fast execution times. It is not good enough when traditional Canny algorithm is applied in complex background images for objects edge detection. It is sensitive to noise and low location accuracy. To solve the shortcomings of traditional Canny algorithm, we propose an improved edge detection and recognition algorithm. The proposed edge detection algorithm can remove the influence of complex background images in the process of edge detection and characters segmentation. The effectiveness of the proposed algorithm is demonstrated by the experimental results. The proposed algorithm can eliminate the complex background of credit card and detect the meaningful edges of its numbers. The experimental results show the improved algorithm can obtain accurate edge position and good segmentation results.

On the basis of Markov Random Field (MRF), which uses context information, in this paper, a robust image segmentation method is proposed. The relationship between observed pixel intensities and distance between pixels are introduced to the traditional neighbourhood potential function, which described the probability of pixels being classified into one class. To perform an unsupervised segmentation, the Bayes Information Criterion (BIC) is used to determine the class number. The K-means is employed to initialise the classification and calculate the mean values and variances of the classes. The segmentation is transformed to maximize a posteriori (MAP) procedure. Then, the Iterative Conditional Model (ICM) is employed to solve the MAP problem. In the experiments, the proposed method is adopted with K-means, traditional Expectation-Maximization (EM) and MRF image segmentation techniques, for noisy image segmentation applying on synthetic and real images. The experiment results and the histogram of signal to noise ratio (SNR)-miss classification ratio (MCR) showed that the proposed algorithm is the better choice.KeywordsImage SegmentationSegmentation ResultMarkov Random FieldUnsupervised SegmentationHouse ImageThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Actively monitoring machine learning models during production operations helps ensure prediction quality and detection and remediation of unexpected or undesired conditions. Monitoring models already deployed in big data environments brings the additional challenges of adding monitoring in parallel to the existing modelling workflow and controlling resource requirements. In this paper, we describe (1) a framework for monitoring machine learning models; and, (2) its implementation for a big data supply chain application. We use our implementation to study drift in model features, predictions, and performance on three real data sets. We compare hypothesis test and information theoretic approaches to drift detection in features and predictions using the Kolmogorov-Smirnov distance and Bhattacharyya coefficient. Results showed that model performance was stable over the evaluation period. Features and predictions showed statistically significant drifts; however, these drifts were not linked to changes in model performance during the time of our study.

A method of real-time object detection using Radial Reach Correlation (RRC) has been developed. We also apply a statistical background estimation to cope with dynamic and complex environments. In order to reduce the computational cost of RRC for an embedded image processing device, we apply two techniques: fast RRC algorithm and background estimation based on statistical approach with cumulative averaging process. As a result, it is possible to reduce the RRC calculation process to about 1/4 in comparison with normal RRC. By combining these techniques, we can achieve robust real-time object detection.

Appropriate statistical analysis of the validation data for diagnostic tests facilitates the evaluation of the performance criteria and increases the confidence in the conclusions drawn from these data. A comprehensive approach to analysing and reporting data from validation studies and inter‐laboratory comparisons such as test performance studies is described. The proposed methods, including statistical analyses, presentation and interpretation of the data, are illustrated using a real dataset generated during a test performance study conducted in the framework of the European project, VALITEST. This analytical approach uses, wherever possible and whenever applicable, statistical analyses recommended by international standards illustrating their application to plant health diagnostic tests. The present work is addressed to plant health diagnosticians and researchers interested and/or involved in the validation of plant diagnostic tests, and also aims to convey the necessary information to those without a statistical background. Detailed statistical explanations are provided in the Appendices.

Background Estimation is a common computer vision task, used for segmenting moving objects in video streams. This can be useful as a pre-processing step, isolating regions of interest for more complicated algorithms performing detection, recognition, and identification tasks, in order to reduce overall computation time. This is especially important in the context of embedded systems like smart cameras, which may need to process images with constrained computational resources. This work focuses on accelerating SuperBE, a superpixel-based background estimation algorithm that was designed for simplicity and reducing computational complexity while maintaining state-of-the-art levels of accuracy. We explore both software and hardware acceleration opportunities, converting the original algorithm into a greyscale, integer-only version, and using Hardware/Software Co-design to develop hardware acceleration components on FPGA fabric that assist a software processor. We achieved a 4.4× speed improvement with the software optimisations alone, and a 2× speed improvement with the hardware optimisations alone. When combined, these led to a 9× speed improvement on a Cyclone V System-on-Chip, delivering almost 38 fps on 320 × 240 resolution images.

Environmental screening of gamma radiation consists of detecting weak nuisance and anomaly signal in the presence of strong and highly varying background. In a typical scenario, a mobile detector-spectrometer continuously measures gamma radiation spectra in short, e.g., one-second, signal acquisition intervals. The measurement data is a 2D matrix, where one dimension is gamma ray energy, and the other dimension is the number of measurements or total time. In principle, gamma radiation sources can be detected and identified from the measured data by their unique spectral lines. Detecting sources from data measured in a search scenario is difficult due to the highly varying background because of naturally occurring radioactive material (NORM), and low signal-to-noise ratio (S/N) of spectral signal measured during one-second acquisition intervals. The objective of this work is to explore unsupervised machine learning (ML) algorithms for development of a digital twin of gamma radiation background, and for detection and identification of weak nuisances and anomalies events in the presence of highly fluctuating background. In one segment of work, we developed a gamma background estimation model using a Longshort term memory (LSTM) network for one-step CPS time series prediction. The LSTM model was validated with two data sets of measurements from two independent NaI detectors positioned on a mobile platform. The data sets contained background radiation only and no orphan isotope sources. The LSTM model was constructed and tested using data from one of the detectors. Performance of the LSTM model was validate through one-step prediction of CPS time series of another NaI detector without re-training. This approach allows to create a digital twin for nuclear background estimation. Using LSTM, it could be possible to detect a source through subtraction of the estimated counts from the measured background. In another segment of work, we investigated detection of gamma emitting sources in the presence of complex background using unsupervised machine learning. Spectral lines of isotopes are difficult to observe in one-second measurements. Averaging over the entire measurement campaign data set reveals spectral lines of most common background isotopes. Spectral lines of orphan sources, which might appear only in a few measurements during the campaign, will be washed out if averaging is performed over the entire measurement data set. The approach we have explored consists of extracting one-second measurements containing weak spectral features through data clustering. Averaging one-second spectra in a cluster should reveal the presence of anomaly sources. We created two ML models using K-means clustering and Neural Network Self-organizing Map (SOM). Performance of these ML models was benchmarked using search data. One data set contained 137Cs source, and another dataset contained 131I source.

Image mosaicing is an effective means of constructing a single panoramic image from a series of snapshots taken in different viewing angles. However, in the case of congested traffic scenes with a cluttered environment including vehicles or pedestrians, there are severe difficulties in aligning a pair of snapshots. In such cases, some objects would be taken only in one of the image pair, thereby resulting in failure in stitching the pair of images. This paper deals with three types of techniques for performing an image mosaicing: Homography estimation for determining geometrical relationships between the image pair, expectation-maximization algorithm for removing inconsistent overlapping region, and graph cuts for seamless stitching and background estimation. Experimental results indicate that the proposed technique is effective to synthesize a panoramic image from a series of narrow-field-of-view snapshots.

Median filter is well-known for researchers in many fields. In image and video processing, it can be used for filtering noises, for estimating background images, and so on. Its effectiveness is undoubtful. However, median filter is time-consuming. This paper presents a new method for computing median filter's response. The proposed method is O(1) and better than existing methods analytically and experimentally. The experimental results with background estimation problem consolidate the finding in this paper.

The paper proposes a dynamic thresholding based image processing technique for the detection of hemorrhages in retinal images. The algorithm uses the information about color and size of hemorrhages as a tool for classifying hemorrhages from other dark lesions present in the retinal images. The algorithm uses the concepts of contrast enhancement, background estimation and intensity variation at edges that is gradient magnitude information supported by some morphological operations. The algorithm follows a simple approach of step by step removal of unwanted features from targeted images using concepts of thresholding and morphology without compromising with accuracy and time of execution. The experimental results indicate that hemorrhages are detected with good accuracy in the retinal images.

Background estimation, i.e. automatic recovery of the background image from a sequence of images containing moving foreground objects, is an important module in many applications, e.g. surveillance and video segmentation. In this paper, we present a simple, yet effective and robust approach for background estimation based on loopy belief propagation. Robustness of the proposed approach means: (i) minimal assumption on the input frames, and (ii) no need to tune parameters. Basically, the background can be recovered even when the occluding foreground objects stay still for a long time. Furthermore, no motion information needs to be known or estimated for the foreground objects, which implies that background can be recovered from a set of frames which are not consecutive temporally. Analysis and experiments are provided to compare the proposed approach to related methods. Experimental results on typical surveillance videos demonstrate the effectiveness of our approach.

Swelling of shale potentially occurs when it is exposed to water-based drilling fluid. The migration of hydrogen ions (H+) in the nano-interlayered platelets of the shale rock is utterly responsible for the swelling behavior in the shale. Conventionally, swelling behavior of any shale formation can be experimentally determined by linear dynamic swell meter. However, it is extremely important to validate these experimental results; hence, this research study aims in conducting a comparative performance analysis for different kinetic models, namely Peleg’s model, first-order exponential association equation and pseudo-second-order kinetic model, and a newly developed scaling swelling model in estimating the experimental results of three different shale samples, namely Talhar, Ranikot and Murree, obtained from different regions of Pakistan. It was found that the performance of the scaling swelling model was the most accurate in predicting the experimental swelling results with accuracy greater than 95% in all the three samples. Peleg’s model is found to be the most inaccurate with $$p \mathrm{values}< \alpha (0.05)$$ p values < α ( 0.05 ) in all the three formations. The equilibrium state in all the three samples was unable to attain by the use of this model. This clearly shows that the transient states continue throughout the course of experimentation, thus demonstrating a higher water activity in the shale samples. Moreover, when comparison was made between the two remaining kinetic adsorption models, it was perceived that pseudo-second-order kinetic was far superior to first-order exponential association equation with $${\mathrm{mean}}_{\mathrm{model}}\simeq {\mathrm{mean}}_{\mathrm{experiment}}$$ mean model ≃ mean experiment and less dispersion in the dataset. Nevertheless, the performance of this model also suffers with the increase in clay content. Furthermore, all these analyses were further validated by different statistical error analysis that includes MAE, APRE% and ANOVA.