Moving object detection has been extensively studied during the last few decades. However the detection of moving objects in different degraded atmospheric conditions (i.e. fog, haze, dust and poor illumination) is less understood. This is possibly because of the lack of a suitable and publically-available video dataset under such weather conditions within which salient objects are unambiguously defined and annotated. This paper describes the creation and design of a new video dataset named as “Tripura University Video dataset (TUVD)” which specifically addresses degraded atmospheric weather conditions for moving object detection in outdoor scenes. The objective is to provide video dataset containing moving objects with annotated ground truth in the form of images of the salient objects in the image sequences. Currently, TUVD contains 55 videos of moving objects (vehicles, animals and pedestrian) under degraded atmospheric conditions. Using TUVD a comparison is made between the results of seven existing state-of-the-art visibility enhancement methods. Quantitative assessment of image quality is achieved using four no-reference image based quality assessment metrics. Overall, the most efficient method for visibility restoration of outdoor scenes is found to be one based on multi-scale fusion, although most of the other algorithms tested show interesting capability in specific cases.

This paper is concerned with the mitigation of simple contrast loss due to added lightness in an image. This added lightness has been referred to as "airlight" in the literature since it is often caused by optical scattering due to fog or mist. A statistical model for scene content is formulated that gives a way of detecting the presence of airlight in an arbitrary image. An algorithm is described for estimating the level of this airlight given the assumption that it is constant throughout the image. This algorithm is based on finding the minimum of a global cost function and is applicable to both monochrome and color images. The method is robust and insensitive to scaling. Once an estimate of airlight is achieved, then image correction is straightforward. The performance of the algorithm is explored using the Monte Carlo simulation with synthetic images under different statistical assumptions. Several examples of before and after color images are given. Results with real video data obtained in poor visibility conditions indicate frame-to-frame consistency of better than 1% of maximum level.

Degradation of images by the atmosphere often restricts imaging applications to good visibility conditions. For example, when imaging the terrain from a forward-looking airborne camera, the atmospheric degradation causes both a loss in contrast and color information. Enhancement of such images is a difficult task due to the complexity in restoring both the luminance and chrominance while maintaining good color fidelity. One particular problem is the fact that the level of contrast loss depends strongly on the wavelength; shorter wavelengths i.e., blue are more effected. In this paper, a novel method is presented for the enhancement of color images. This method is based on the underlying physics of the degradation and the parameters required for enhancement are estimated from the image itself. The proposed method is tested using synthetic images to explore the limitations and reliability of the method under different visibility conditions. Enhancement is performed on real images taken using an airborne camera at a height of approximately 1000 meters in hazy conditions for which the visibility is approximately 10 kilometers. Significant improvements in terms of contrast, visible range and color fidelity are evident when compared to existing methods.

In daylight viewing conditions, image contrast is often significantly degraded by atmospheric aerosols such as haze and fog. This paper introduces a method for reducing this degradation in situations in which the scene geometry is known. Contrast is lost because light is scattered toward the sensor by the aerosol particles and because the light reflected by the terrain is attenuated by the aerosol. This degradation is approximately characterized by a simple, physically based model with three parameters. The method involves two steps: first, an inverse problem is solved in order to recover the three model parameters; then, for each pixel, the relative contributions of scattered and reflected flux are estimated. The estimated scatter contribution is simply subtracted from the pixel value and the remainder is scaled to compensate for aerosol attenuation. This paper describes the image processing algorithm and presents an analysis of the signal-to-noise ratio (SNR) in the resulting enhanced image. This analysis shows that the SNR decreases exponentially with range. A temporal filter structure is proposed to solve this problem. Results are presented for two image sequences taken from an airborne camera in hazy conditions and one sequence in clear conditions. A satisfactory agreement between the model and the experimental data is shown for the haze conditions. A significant improvement in image quality is demonstrated when using the contrast enhancement algorithm in conjuction with a temporal filter.

This paper is concerned with the implementation of oriented Gaussiantype filters with several possible decomposition schemes, including the steerable approach and the wavelet transform. A novel spatial decomposition scheme is introduced, in which the filters are projected onto a set of Gaussian basis functions which are not dependent on the orientation of the filters. This scheme only costs a few multiply/add operations per pixel per orientation and provides a considerable improvement in efficiency. 1

Intelligent Document Understanding (IDU) is the process of converting scanned document pages into an electronic, processable form. We have previously presented a IDU system architecture suitable for this task which uses a hybrid bottom-up/top-down control strategy. In this paper we focus on a specific subproblem that arises within the chosen framework, concerned with selecting an appropriate page layout structure. A detailed analysis of the problem using an error propagation model, allows computationally simple search strategies to be developed. A multistage layout formation algorithm is proposed and its performance is critically assessed when implemented using two different Layout Object selection criterion. The first selection criterion is based on a maximal column area coverage; the second is based on a probabilistic Layout Object selection. Both techniques have been incorporated into the hybrid IDU system and the results presented indicate its superiority over previously reported systems.

International audience

The paper presents a system for the conversion of scanned documents into the open document architecture. Unlike previous work in this field the authors use a combination of evidence sources to achieve greater robustness to document defects and noise introduced in the scanning process. Furthermore, they use optical character recognition in conjunction with other forms of image analysis as a means of detecting document structure. This enables enhanced document feature extraction and improved performance. They demonstrate the performance of the system on a specific class of input document.>

A novel method for the interpolation of sampled images is presented. It makes use of a recently discovered formula for the least-squares projection of an arbitrary function onto a repetitive basis. The proposed interpolation formula differs from standard techniques such as cubic spline convolution in that the image samples are modified by a discrete convolution operator prior to the reconstruction summation. The visual performance of the method is shown to be superior to that of cubic spline convolution, which is the best current algorithm. The main attraction of the method is that the algorithm is automatically tailored to the spatial resolution of the image sensor. The exact computational cost of the method, in terms of reconstruction sum size, depends on the sensor PSF (point spread function) but is likely to be only slightly greater than that for spline convolution. All the results given hold good in a general N-dimensional space.>