To improve the localization capability of the mobile robot in a crowded and unorderly indoor environment, an approach for extracting LSRII (Local Salient Region Integral Invariant) features is ...proposed. The approach extracts integral invariant features in salient regions in an image. The global localization is achieved by applying LSRII features in particle filter localization. The practical experiments illustrate that our approach is reliable in a crowded and unorderly indoor environment.
With the six-electrode probe, a two-phase flow measurement system was designed for oil well logging. The whole system consists of the excitation signal generating circuit, the signal conditioning ...circuit, the data sampling & processing circuit. This apparatus can run into oil production well and interface to a PC on the ground via RS485 bus. Hybrid processor MC56F8323 was the control center of the proposed system. It sampled the up-stream and down-stream fluid flowing signals and cross-correlated in frequency-domain for axial velocity measurement. Simultaneously, MC56F8323 sampled the distributed information of oil-water two-phase flow and estimated the normalized conductivity which is a function of water volume fraction. Especially, the probe was excited by the bidirectional pulse constant-current source, and the traditional complex AC circuit such as demodulation wasn't required, so the circuit simplified and the measurement accuracy improved. The experiments, which were performed in vertical upward oil-water two-phase flow loop, indicate that the measurement error of water volume fraction is plusmn3% when the flowrate of oil-water mixture exceeds 10 m 3 /d, and the flow axial velocity can resolve the flowrate difference with 2 m 3 /d.
The performance of ring-shape electrodes for measuring the conductance of oil-in-water mixtures in pipes was studied theoretically and experimentally. In particular the work is aimed at optimizing ...the probe geometry in order to decrease the axial current leakage and improve the measuring precision of water fraction under oil-well logging application. Starting from the basic four-electrode probe including two exciting and two measuring rings, a new six-electrode probe characterized by adding two guarding rings to be found, the distance and width of guarding electrodes were optimized theoretically. The experiment was performed with reference to annular, stratified and bubble phase distributions and the results were compared with available theoretical models. The analysis demonstrates the response of new six-electrode probe is accurate and reliability, the measuring precision is enhanced.
A novel method is proposed in this paper to detect the types of quadric surface models. Firstly, using high-precision 3D scanner, point clouds are acquired from quadric surface models. Secondly, ...triangulated irregular network models are generated from the point clouds by Delaunay algorithm. Thirdly, normal vectors of the nearest neighborhood planes of points in point clouds are acquired by the least square fitting algorithm. Finally, mapping relationship between the distributions of normal vectors and the types of quadric surface models is constructed. As a result, in real applications, if the distribution of normal vectors obtained for a certain scanned quadric surface model is consistent with the constructed mapping relationship, surface type of the model can be rapidly identified. The proposed method is validated by using an experiment.
Image denoising is a significant inverse problem of image processing and an important image pretreatment. The performance of image denoising is improved by using some statistic characteristics of ...natural image. In this paper, we combine the extensive self-similarity of images in non-local means algorithm with the minimum mean square error of Wiener filtering in wavelet domain, and then propose an image denoising algorithm based on the non-local means with Wiener filtering in wavelet domain. The experimental results demonstrate that one can get denoised image with higher subjective visual quality and peak signal to noise ratio based on the proposed algorithm.
U umreženom robotskom sustavu, robot i senzorska mreža su međusobno integrirani i povezani na način da i jedan i drugi iskoriste svoje prednosti, te da imaju koristi jedan od drugoga. Kako bi ...umreženi robotski sustav mogao djelovati u radnom okruženju potrebno je riješiti tri međusobno povezana problema: lokalizaciju, kalibraciju senzorske mreže i kartiranje prostora. U ovom radu razmatraju se problemi istodobne lokalizacije, kalibracije i kartiranja te se razmatraju mogućnosti poboljšanja njihove preciznosti. Povezanost lokalizacije, kartiranja i kalibracije predstavljena je pomoću zajedničke uvjetne razdiobe i zatim rastavljena u tri razdvojena analitička izraza korištenjem Bayesovih i Markovljevih svojstava. Za rješavanje svih triju analitičkih izraza koristi se Rao-Blackwell čestično filtriranje, pri čemu se prošireni čestični filtar koristi kod lokalizacije a nederivirajući Kalmanov filtar za kalibraciju i kartiranje. Ispravnost i efikasnost predloženog pristupa pokazana je kroz provedene simulacije.