Xin Luan

Illumination invariant color model for object recognition in robot soccer

In this paper, a new illumination invariant color model HSy with large discriminating power is proposed for object recognition in autonomous robot soccer. Object recognition for soccer robot is used as a benchmark problem for the proposed color model. Classical illumination invariant color components are analyzed. New illumination invariant color components are defined based on the dichromatic reflection model. A new color model HSy is constructed by selecting the less correlated components invariant to illumination. The proposed color model is invariant to illumination intensity in the context of the autonomous robot soccer and it yields maximal color discrimination for object recognition. Experiments have been carried out on the soccer robot M-TR for the ball recognition. Experimental results show that the proposed model achieves high object recognition accuracy.

Illumination-robust area-based stereo matching with improved census transform

This paper presents a novel area-based stereo matching algorithm based on improved census transform under changing illumination. For the traditional census-based stereo matching, the result is not robust under variant illumination, because the intensity value of center pixel in the mask is affected by the noise to cause distortion. In order to solve this problem, we propose the improved census transform method, which takes the standard deviation of census mask as the base point instead of the center pixel, comparing with the difference of per neighborhood pixel and the mean intensity of the mask to build the sparse census transform. The experiments show that the stereo matching algorithm is robust even if the illumination changes.

A Novel High Accuracy Sub-Pixel Corner Detection Algorithm for Camera Calibration

This paper presents a novel sub-pixel corner detection algorithm for camera calibration. In order to achieve high accuracy and robust performance, the pixel level candidate regions are firstly identified by Harris detector. Within these regions, the center of gravity (COG) method is used to gain sub-pixel corner detection. Instead of using the intensity value of the regions, we propose to use corner response function (CRF) as the distribution of the weights of COG. The results of camera calibration experiments show that the proposed algorithm is more accurate and robust than traditional COG sub-pixel corner detection methods.

An algorithm of calculating turbulence kinetic energy dissipation rate based on motion compensation

In order to maximum eliminate the pollution caused by vibration of the instrument and improve the accuracy of the turbulent kinetic energy dissipation rate, this paper puts forward an algorithm based on motion compensation. Firstly, spectrum analysis is used to analyze the relationship between shear signal and the acceleration signal. Then, according to the relationship, the acceleration signal takes as reference signal to design motion compensation algorithm. Finally, wiener filtering is used to get the optimum parameter to eliminate instrument vibration interference. In order to verify the validity of the algorithm, real sea data is applied, and the results show that the vibration signal is effectively eliminated, which provides real and effective turbulence signal to calculate the turbulent kinetic energy dissipation rate.

An Improved x-Corners Detection Algorithm on Distorted Images

The accuracy and robustness of x-corners detection are critical in camera calibration. In order to achieve high accuracy and robust performance in blurred and heavily distorted images automatically, the Martin Rufli’s algorithm is improved. The Martin Rufli’s algorithm is robust to detect x-corners in blurred and heavily distorted images; however, the x-corners detected are not accurate enough, and so we take the center of gravity, which is weighted by the squared gradients in order to refine x-corners. As gradient represents grayscale change between a pixel and its adjacent pixels, more accurate x-corners can be detected. Compared with the method proposed by Martin Rufli, experiment results indicate that the improved algorithm can improve accuracy capacity.

The Research about No-Vibration Data Storage of Micro-Scale and Long-Term Ocean Turbulence Measurement

Turbulence played an important role in the evolution of the seawater energy and exchange. Multi-scale, long-term, fixed-point and continuous sampling is a new research direction in the turbulence observation. This dissertation designed high-capacity and no-vibration data storage solutions aiming at long-term, continuous turbulence observations. First a multi-scale submerged buoy observing platform is designed. Base on the turbulence observing platform, a multi-parameter data acquisition and no vibrations storage system is designed. This paper describes the hardware and software design implementation of large-capacity data storage arrays in details as well as the readability and easy operation of the transplant of FatFS. Actual test and sea trial prove the design can be achieved large-capacity data access of long-term observation of ocean turbulence base on the submerged buoy.

An Algorithm of Turbulence Wavenumber Spectrum Matching Based on SVM

Considering the problem of invalid data caused mismatch of wavenumber spectrum which contained in turbulence observation data, an algorithm of turbulent wavenumber spectrum matching based on SVM is proposed. Category labels are obtained from pre-processed raw data by cross validation algorithm, and then the optimum parameters of the classifier are got through SVM learning algorithm. Sea trial data validation results indicate that the algorithm has high matching accuracy, and provides a new way to calculate the turbulence wavenumber spectrum matching.

On-Line Adaptation to Illumination Change for Mobile Robot Based on Omni-Directional Vision

In this paper a method of on-line adaptation to illumination is proposed for mobile robot based on omni-directional in a changing illumination environment. Illumination condition is represented by an average luminance distribution of a reference object in a time series images. Illumination change is detected by computing the KL-divergence between two different distributions. A dual-threshold strategy is used to classify the current illumination into known conditions or an unknown one. According to illumination the robot decides to switch to a corresponding color calibration or learn a new one. Experiments have been carried out on the soccer robot M-TR. Experimental results show the efficiency of the proposed method.

Research on Color Feature Selection Using Genetic Algorithm in Robot Soccer

Feature selection is a hot topic in the field of pattern recognition. In this paper, we present a new feature selection algorithm which is used on the soccer robot MT-R for the ball recognition. The illumination invariant color feature set is defined based on the dichromatic reflection. By means of genetic algorithm we determine the most discriminating color feature subset. Experimental results show that the proposed color feature subset achieves high object recognition accuracy.

A target localization method for soccer robot based on Omni-directional vision

Target localization for soccer robot based on Omni-directional vision is considered. The vision system of the MT-R robot is described. For Omni-directional vision, the target is recognized from the image through the threshold value division method. The corresponding relationship of the target position between the image coordinate and the robot coordinate is analyzed. The method of image polar coordinate transforming and piece-wise defined Lagrange interpolation for target location is adopted. Experiments have been carried out on the soccer robot M-TR for the ball recognition. Experimental results show the effectiveness of the proposed localization method.