Pavel Zemcik

Compression of multispectral remote sensing images using clustering and spectral reduction

Image compression has been one of the main research topics in the field of image processing for a long time. The research usually focuses on compressing images that are visible to humans. The images being compressed are usually gray-level images or RGB color images. Recent advances in technology, however, enable the authors to make the detailed processing of spectral features in the images. Therefore, the compression of images with many spectral channels, called multispectral images, is required. Many methods used in traditional lossy image compression can be reused also in the compression of multispectral images. In this paper, a new combination of clustering spectra, manipulating spectral vectors, and encoding and decoding for multispectral images is presented. In the manipulation of the spectral vectors PCA, ICA, and wavelets are used. The approach is based on extracting relevant spectral information. Furthermore, some quantitative quality measures for multispectral images are presented.

Convolutional Neural Networks for Direct Text Deblurring.

In this work we address the problem of blind deconvolution and denoising. We focus on restoration of text documents and we show that this type of highly structured data can be successfully restored by a convolutional neural network. The networks are trained to reconstruct high-quality images directly from blurry inputs without assuming any specific blur and noise models. We demonstrate the performance of the convolutional networks on a large set of text documents and on a combination of realistic de-focus and camera shake blur kernels. On this artificial data, the convolutional networks significantly outperform existing blind deconvolution methods, including those optimized for text, in terms of image quality and OCR accuracy. In fact, the networks outperform even state-of-the-art non-blind methods for anything but the lowest noise levels. The approach is validated on real photos taken by various devices.

Local Rank Patterns --- Novel Features for Rapid Object Detection

This paper presents Local Rank Patterns (LRP) - novel features for rapid object detection in images which are based on existing features Local Rank Differences (LRD). The performance of the novel features is thoroughly tested on frontal face detection task and it is compared to the performance of the LRD and the traditionally used Haar-like features. The results show that the LRP surpass the LRD and the Haar-like features in the precision of detection and also in the average number of features needed for classification. Considering recent successful and efficient implementations of LRD on CPU, GPU and FPGA, the results suggest that LRP are good choice for object detection and that they could replace the Haar-like features in some applications in the future.

Hardware acceleration of graphics and imaging algorithms using FPGAs

Computer graphics algorithms and algorithms used in image processing are generally computationally expensive. This fact is the reason why people struggle to accelerate such algorithms using any reasonable means. The traditional sources of speedup are faster processors, parallelism, or dedicated hardware. Development in digital circuit technology, especially rapid development of Field Programmable Gate Arrays (FPGA), offers alternative way to acceleration. Current FPGA chips are capable of running graphics algorithms at the speed comparable to dedicated graphics chips. At the same time they are configurable not only using schematics diagram but also through high-level programming languages, e.g. VHDL. The contribution addresses these issues, general development in the area, and shows examples of hardware platforms and algorithms that can be implemented on such platforms.

Incremental Block Cholesky Factorization for Nonlinear Least Squares in Robotics

Efficiently solving nonlinear least squares (NLS) problems is crucial for many applications in robotics. In online applications, solving the associated nolinear systems every step may become very expensive. This paper introduces online, incremental solutions, which take full advantage of the sparseblock structure of the problems in robotics. In general, the solution of the nonlinear system is approximated by incrementally solving a series of linearized problems. The most computationally demanding part is to assemble and solve the linearized system at each iteration. In our solution, this is mitigated by incrementally updating the factorized form of the linear system and changing the linearization point only if needed. The incremental updates are done using a resumed factorization only on the parts affected by the new information added to the system at every step. The sparsity of the factorized form directly affects the efficiency. In order to obtain an incremental factorization with persistent reduced fill-in, a new incremental ordering scheme is proposed. Furthermore, the implementation exploits the block structure of the problems and offers efficient solutions to manipulate block matrices, including a highly efficient Cholesky factorization on sparse block matrices. In this work, we focus our efforts on testing the method on SLAM applications, but the applicability of the technique remains general. The experimental results show that our implementation outperforms the state of the art SLAM implementations on all the tested datasets.

Efficient implementation for block matrix operations for nonlinear least squares problems in robotic applications

A large number of robotic, computer vision and computer graphics applications rely on efficiently solving the associated sparse linear systems. Simultaneous localization and mapping (SLAM), structure from motion (SfM), non-rigid shape recovery, and elastodynamic simulations are only few examples in this direction. In general, these problems are nonlinear and the solution can be approximated by incrementally solving a series of linearized problems. In some applications, the size of the system considerably affects the performance, especially when the sparsity is low. This paper exploits the block structure of such problems and offers very efficient solutions to manipulate block matrices within iterative nonlinear solvers. The resulting method considerably speeds-up the execution of the implementation of the nonlinear optimization problem. In this work, in particular, we focus our effort on testing the method on SLAM applications, but the applicability of the technique remains general. Our implementation outperforms the state of the art SLAM implementations on all tested datasets. In incremental mode, where a larger portion of time is spent in updating the system, our implementation is on average two times faster than the others.

Particle rendering pipeline

This paper presents a particle rasterizing pipeline that is intended as a part of a particle system rendering engine. The purpose of the rasterizing pipeline is to convert the particles, which are geometrically just a projection of circles onto a plane, into pixels of a raster image. While the conversion is relatively simple, the required speed is very high as the particle systems typically contain very large numbers of particles - at least hundreds of thousands - and the general goal is to handle the rendering in real time. The presented solution does have the capability of achieving such high speeds.

Feature point detection under extreme lighting conditions

This paper evaluates the suitability of High Dynamic Range (HDR) imaging techniques for feature point detection under extreme lighting conditions. The conditions are extreme in respect to the dynamic range of the lighting within the test scenes used. This dynamic range cannot be captured using standard low dynamic range imagery techniques without loss of detail. Four widely used feature point detectors are used in the experiments: Harris corner detector, Shi-Tomasi, FAST and Fast Hessian. Their repeatability rate is studied under changes of camera viewpoint, camera distance and scene lighting with respect to the image formats used. The results of the experiments show that HDR imaging techniques improve the repeatability rate of feature point detectors significantly.

Fast covariance recovery in incremental nonlinear least square solvers

Many estimation problems in robotics rely on efficiently solving nonlinear least squares (NLS). For example, it is well known that the simultaneous localisation and mapping (SLAM) problem can be formulated as a maximum likelihood estimation (MLE) and solved using NLS, yielding a mean state vector. However, for many applications recovering only the mean vector is not enough. Data association, active decisions, next best view, are only few of the applications that require fast state covariance recovery. The problem is not simple since, in general, the covariance is obtained by inverting the system matrix and the result is dense. The main contribution of this paper is a novel algorithm for fast incremental covariance update, complemented by a highly efficient implementation of the covariance recovery. This combination yields to two orders of magnitude reduction in computation time, compared to the other state of the art solutions. The proposed algorithm is applicable to any NLS solver implementation, and does not depend on incremental strategies described in our previous papers, which are not a subject of this paper.

A comparative study on distant free-hand pointing

In this paper we present a comparative study of free-hand pointing, an absolute remote pointing device. Unimanual and bimanual interaction were tested as well as the static reference system (spatial coordinates are fixed in the space in front of the TV) and novel body-aligned reference system (coordinates are bound to the current position of the user). We conducted a point-and-click experiment with 12 participants. We have identified the preferred interaction areas for left- and right-handed users in terms of hand preference and preferred spatial areas of the interaction. In bimanual interaction, the users relied more on dominant hand, switching hands only when necessary. Even though the remote pointing device was faster than the free-hand pointing, it was less accepted probably due to its low precision.