Blind source separation for convolutive mixtures based on the joint diagonalization of power spectral density matrices

This paper studies the problem of blind separation of convolutively mixed source signals on the basis of the joint diagonalization (JD) of power spectral density matrices (PSDMs) observed at the output of the separation system. Firstly, a general framework of JD-based blind source separation (BSS) is reviewed and summarized. Special emphasis is put on the separability conditions of sources and mixing system. Secondly, the JD-based BSS is generalized to the separation of convolutive mixtures. The definition of a time and frequency dependent characteristic matrix of sources allows us to state the conditions under which the separation of convolutive mixtures is possible. Lastly, a frequency-domain approach is proposed for convolutive mixture separation. The proposed approach exploits objective functions based on a set of PSDMs. These objective functions are defined in the frequency domain, but are jointly optimized with respect to the time-domain coefficients of the unmixing system. The local permutation ambiguity problems, which are inherent to most frequency-domain approaches, are effectively avoided with the proposed algorithm. Simulation results show that the proposed algorithm is valid for the separation of both simulated and real-word recorded convolutive mixtures.     

Three-dimensional profilometry based on shift estimation of projected fringe patterns

This paper presents a new approach to fringe pattern profilometry. In this paper, a generalized model describing the relationship between the projected fringe pattern and the deformed fringe pattern is derived, in which the projected fringe pattern can be arbitrary rather than being limited to being sinusoidal, as are those for the conventional approaches. Based on this model, what is believed to be a new approach is proposed to reconstruct the three-dimensional object surface by estimating the shift between the projected and deformed fringe patterns. Additionally, theoretical analysis, computer simulation, and experimental results are presented, which show how the proposed approach can significantly improve the measurement accuracy, especially when the fringe patterns are distorted by unknown factors.     

A roll eccentricity sensor for steel-strip rolling mills

The problem of roll eccentricity compensation in steel-strip rolling mills is examined. A novel roll eccentricity sensor is proposed for the task of retrieving the desired reference signal from the available roll-force measurement. The sensor implements a modified comb filter structure which can be arranged in either tunable or adaptive form. Simulation results based on typical experimental data indicate that the proposed sensor is appropriate and offers distinct advantages over other methods. The proposed comb filter is new in the sense that it consists of a number of second-order notch filter models with specially constrained bandwidth characteristics. An approximate gradient-based adaptive algorithm is used for the frequency estimation. The important feature of such an algorithm is that it is robust and is simple to implement. The advantages of the sensor include fast convergence and accurate parameter estimates. No extra positional sensors are required since this information can be easily obtained from the phase estimates. Finally, the sensor is not based on approximate mill models; it treats the problem as a harmonic series buried in noise. The resulting solution is based on time domain measurements that characterize the mill roll eccentricity conditions during operation     

Toward Automatic Measurement of the Linewidth-Enhancement Factor Using Optical Feedback Self-Mixing Interferometry With Weak Optical Feedback

This paper proposes an approach for automatically measuring the linewidth-enhancement factor (LEF) of semiconductor lasers using optical feedback self-mixing interferometry (OFSMI), which works in weak optical feedback regime and where the external target is subject to simple harmonic vibration with unknown vibration frequency and magnitude. According to well-known Lang-Kobayashi theory the waveform of the modulated optical output power from the OFSMI system is influenced by multiple parameters, including the LEF, the optical feedback level factor, and the parameters related to the movement of external target. In order to estimate LEF, other parameters must also be considered and, hence, a multiple parameter estimation strategy is required. We propose a solution for this multiple parameter estimation problem based on the principle of data-to-theoretical model match. In particular, a strategy for minimizing a cost function in order to achieve the best fitting is proposed with which all the unknown parameters can be estimated. The performance of the proposed approach is tested using experimental data in comparison with other two approaches. It is seen that, over different experimental signals, the standard deviation for estimated LEF is less than 4.58% on average, which shows that results have excellent consistency. Moreover, the proposed approach also provides a solution for vibration measurement (that is, vibration frequency and magnitude).     

Blind color isolation for color-channel-based fringe pattern profilometry using digital projection

We present an algorithm for estimating the color demixing matrix based on the color fringe patterns captured from the reference plane or the surface of the object. The advantage of this algorithm is that it is a blind approach to calculating the demixing matrix in the sense that no extra images are required for color calibration before performing profile measurement. Simulation and experimental results convince us that the proposed algorithm can significantly reduce the influence of the color cross talk and at the same time improve the measurement accuracy of the color-channel-based phase-shifting profilometry.     

Discrete cosine transform-based shift estimation for fringe pattern profilometry using a generalized analysis model

What is believed to be a new analysis algorithm to carry out profile measurement with low computational complexity and less noise sensitivity is presented. First, a discrete cosine transform (DCT)-based representation method is introduced to express the height distribution of a 3D surface. Then a novel shift estimation algorithm, called the DCT-based shift estimation (DCT-SE), is presented to reconstruct 3D object surfaces by using the proposed expression and the generalized analysis model. The advantage of DCT-SE is that without loss of measurement precision it provides lower computational complexity to implement 3D reconstruction from nonlinearly distorted fringe patterns and, at the same time, survives the random noise. Simulations and experiments show that the proposed DCT-SE is a fast, accurate, and efficient reconstruction algorithm for digital projection- based fringe pattern profilometry techniques.