Evolutionary RBF classifier for polarimetric SAR images

In this paper, a robust radial basis function (RBF) network based classifier is proposed for polarimetric synthetic aperture radar (SAR) images. The proposed feature extraction process utilizes the covariance matrix elements, the H/α/A decomposition based features combined with the backscattering power (span), and the gray level co-occurrence matrix (GLCM) based texture features, which are projected onto a lower dimensional feature space using principal components analysis. For the classifier training, both conventional backpropagation (BP) and multidimensional particle swarm optimization (MD-PSO) based dynamic clustering are explored. By combining complete polarimetric covariance matrix and eigenvalue decomposition based pixel values with textural information (contrast, correlation, energy, and homogeneity) in the feature set, and employing automated evolutionary RBF classifier for the pattern recognition unit, the overall classification performance is shown to be significantly improved. An experimental study is performed using the fully polarimetric San Francisco Bay and Flevoland data sets acquired by the NASA/Jet Propulsion Laboratory Airborne SAR (AIRSAR) at L-band to evaluate the performance of the proposed classifier. Classification results (in terms of confusion matrix, overall accuracy and classification map) compared with the major state of the art algorithms demonstrate the effectiveness of the proposed RBF network classifier.     

On root structures and convergence properties of weighted median filters

A weighted median filter is a nonlinear digital filter consisting of a window of length 2N + 1 and a weight vector W=(W_? N,..., W₀,..., W_N). A root signal of a median type filter is a signal that is invariant to the filter. However, not all weighted median filters possess the convergence property. In this paper, we shall study the root structures and the convergence behavior of a subclass of weighted median filters, calledclass- 1 filters, which is symmetric in its weight vector. We shall introduce an important parameter, calledfeature value, and show that any one-dimensional unappended signal of lengthL will converge to a root signal in at most $$3\left\lceil {\frac{{L - 2}}{{2(2N + 2 - p)}}} \right\rceil$$ passes of aclass ?1 filter with window width 2N + 1 and thefeature value p.     

Modeling the timing of cuts in automatic editing of concert videos

Multimedia Tools and Applications - Increasing amount of video content is being recorded by people in public events. However, the editing of such videos can be challenging for the average user. We...     

Multimodal extraction of events and of information about the recording activity in user generated videos

In this work we propose methods that exploit context sensor data modalities for the task of detecting interesting events and extracting high-level contextual information about the recording activity in user generated videos. Indeed, most camera-enabled electronic devices contain various auxiliary sensors such as accelerometers, compasses, GPS receivers, etc. Data captured by these sensors during the media acquisition have already been used to limit camera degradations such as shake and also to provide some basic tagging information such as the location. However, exploiting the sensor-recordings modality for subsequent higher-level information extraction such as interesting events has been a subject of rather limited research, further constrained to specialized acquisition setups. In this work, we show how these sensor modalities allow inferring information (camera movements, content degradations) about each individual video recording. In addition, we consider a multi-camera scenario, where multiple user generated recordings of a common scene (e.g., music concerts) are available. For this kind of scenarios we jointly analyze these multiple video recordings and their associated sensor modalities in order to extract higher-level semantics of the recorded media: based on the orientation of cameras we identify the region of interest of the recorded scene, by exploiting correlation in the motion of different cameras we detect generic interesting events and estimate their relative position. Furthermore, by analyzing also the audio content captured by multiple users we detect more specific interesting events. We show that the proposed multimodal analysis methods perform well on various recordings obtained in real live music performances.     

Isolated regions in video coding

Different types of prediction are applied in modern video coding. While predictive coding improves compression efficiency, the propagation of transmission errors becomes more likely. In addition, predictive coding brings difficulties to other aspects of video coding, including random access, parallel processing, and scalability. In order to combat the negative effects, video coding schemes introduce mechanisms such as slices and intracoding, to limit and break the prediction. This paper proposes the use of the isolated regions coding tool that jointly limits in-picture prediction and interprediction on a region-of-interest basis. The tool can be used to provide random access points from non-intrapictures and to respond to intrapicture update requests. Furthermore, it can be applied as an error-robust macroblock mode decision method and can be used in combination with unequal error protection. Finally, it enables mixing of scenes, which is useful in coding of masked scene transitions.     

Forecasting Financial Time Series Using Robust Deep Adaptive Input Normalization

Deep Learning provided powerful tools for forecasting financial time series data. However, despite the success of these approaches on many challenging financial forecasting tasks, it is not always straightforward to employ DL-based approaches for highly volatile and non-stationary time financial series. To this end, in this paper, an adaptive input normalization layer that can learn to identify the distribution from which the input data were generated and then apply the most appropriate normalization scheme is proposed. This allows for promptly adapting the input to the subsequent DL model, which can be especially important, given recent findings that hint at the existence of critical learning periods in neural networks. Furthermore, the proposed method operates on a sliding window over the time series allowing for overcoming non-stationary issues that often arise. It is worth noting that the main difference with existing approaches is that the proposed method does not just learn to perform static normalization, e.g., using a fixed set of parameters, but instead it adaptively calculates the most appropriate normalization parameters, significantly improving the robustness of the proposed approach when distribution shifts occur. The effectiveness of the proposed formulation is verified using extensive experiments on three challenging financial time-series datasets.

     

Parallel watershed transformation algorithms for image segmentation

The watershed transformation is a mid-level operation used in morphological image segmentation. Techniques applied on large images, which must often complete fast, are usually computationally expensive and complex entailing efficient parallel algorithms. Two distributed approaches of the watershed transformation are introduced in this paper. The algorithms survey in a Single Program Multiple Data (SPMD) model both local and global connectivity properties of the morphological gradient of a gray-scale image to label connected components. The sequentiality of the serial algorithm is broken in the parallel versions by exploiting the ordering relation between two neighboring pixels successively incorporated in the same region. Thus, a path is traced, for every unlabeled pixel, down to its region of inclusion (whose label is then propagated backwards); in the second algorithm, regions grow independently around their seeds. In both cases only pixels which satisfy the ordering relation are incorporated in any region. This way, not only different regions are explored in a parallel fashion, but also different parts of the same region, when the latter extends to neighboring subdomains, are treated likewise. Running time and relative speedup evaluated on a Cray T3D parallel computer are used to appreciate the performance of both algorithms.     

Analysis of two-dimensional center weighted median filters

Center weighted median (CWM) filters, which have been recognized as detail preserving filters, are an important and the simplest subclass of weighted median (WM) filters. In this paper, we analyze the root signals of two-dimensional (2-D) CWM filters. In particular, we derive the required form for a signal to be a root of a 2-D CWM filter. The required form of signals to be roots is then used to evaluate the detail preserving properties of 2-D CWM filters. As examples, the detail preserving properties of some 2-D CWM filters are compared with other detail preserving filters, i.e. multilevel median filters. The generation of binary root signals of some 2-D CWM filters is treated in the term of the smallest surviving object (SSO). It is illustrated by some examples that CWM filters with different orientation of windows can be useful in image segmentation.     

An effective CU size decision method for quality scalability in SHVC

The Scalable extension of the High Efficiency Video Coding (known as SHVC) combines the high compression efficiency with the possibility of encoding different resolutions of the same encoded video in a single bitstream. However, this is accompanied with a high computational complexity. In this paper, we propose an effective coding unit (CU) size decision method by restricting the CU depth range to reduce the encoding time for quality scalability in SHVC. Since the optimal depth level in the enhancement layer (EL) is highly correlated to that in the base layer (BL), we can determine the CU depth range in the EL according to the depth of the co-located CU in the BL. Based on the high correlation between the current CU and its spatio-temporal neighboring CUs, the proposed method skips some specific depth levels which are rarely used in the previous frame and neighboring CUs to further reduce the computational complexity. Experimental results demonstrate that the proposed method can efficiently reduce computational complexity while maintaining similar rate distortion (RD) performance as the original SHVC encoder.     

An overview of median and stack filtering

Within the last two decades a small group of researchers has built a useful, nontrivial theory of nonlinear signal processing around the median-related filters known as rank-order filters, order-statistic filters, weighted median filters, and stack filters. This required significant effort to overcome the bias, both in education and research, toward linear theory, which has been dominant since the days of Fourier, Laplace, and Convolute.We trace the development of this theory of nonlinear filtering from its beginnings in the study of noise-removal properties and structural behavior of the median filter to the recently developed theory of optimal stack filtering.The theory of stack filtering provides a point of view which unifies many different filter classes, including morphological filters, so it is discussed in detail. Of particular importance is the way this theory has brought together, in a single analytical framework, both the estimation-based and the structural-based approaches to the design of these filters.Some recent applications of median and stack filters are provided to demonstrate the effectiveness of this approach to nonlinear filtering. They include: the design of an optimal stack filter for image restoration; the use of vector median filters to attenuate impulsive noise in color images and to eliminate cross luminance and cross color in TV images; and the use of median-based filters for image sequence coding, reconstruction, and scan rate conversion in normal TV and HDTV systems.