Guoguang Mu

Deconvolution methods for image deblurring in optical coherence tomography

Two-dimensional deconvolution methods are proposed to deblur optical coherence tomography images. One employs a two-dimensional deconvolution with a matrix given by the product of the longitudinal and transversal point-spread functions as its kernel, which can be taken as the general point-spread function of an optical coherence tomography system. The other uses two one-dimensional deconvolutions with the longitudinal and transversal point-spread functions successively. It is shown that the two deconvolution methods can deblur the experimentally obtained optical coherence tomography images effectively.

Improving binary images reconstructed from kinoforms by amplitude adjustment

Abstract A method for amplitude adjustment is proposed in this paper for the first time to improve the quality of binary images obtained from kinoforms by Iterative Fourier Transform Algorithm (IFTA). By setting the amplitude of the “zero” area in a binary image to a low but non-zero amplitude, their phase is involved in the iteration, which results in an increase of the freedom degree in the design of kinoforms, and an improvement of the quality of the reconstructed binary images. Quantitative illustrations and discussions of the improvement effect are given from simulation results.

Nonlinear imaging of embedded microstructures inside transparent materials with laser-induced ionization microscopy

We present a novel nonlinear imaging method that utilizes femtosecond laser-induced plasma emission to probe microscopic structures embedded inside transparent materials. This nonlinear diagnostic tool can resolve either elemental or structural variations of the sample of interest and provide significant improvements over the ordinary linear microscopes by having much higher contrast ratios for the observed areas of different refractive indices. Examples of using this technique to examine the microstructures fabricated by ultrashort laser pulses inside optical glass are presented.

3D monochromatic image synthesized with vertical area-partitioned recording of master hologram in multiple-exposure holography

We report for the first time theoretical analysis and experimental results of a vertical-area partition method for recording master holograms in multiple-exposure rainbow holography to synthesize monochromatic 3D image from a series of medical tomograms. In this novel method, the master hologram is area partitioned into elementary master holograms, which are recorded in a periodic arrangement along the vertical direction. Under the white-light reconstruction, a 3D monochromatic image composed of a series of medical tomograms can be synthesized with wide viewing angle, high resolution, and low color blur.

Ultrashort pulse laser ionization microscopy

Fundamental principles and advantages of the laser induced air ionization microcopy (LIAIM) and laser induced breakdown microscopy (LIBM) are introduced. Potential applications of these two new types of nonlinear imaging methods using ultrashort laser pulses in imaging both dielectric materials and bio-samples are demonstrated with some representative experimental results. Effects of different laser pulse widths on the discrimination power for laser-written microstructures inside transparent materials and the elemental composition are also investigated. It is shown that femtosecond laser induced ionization probe detects the variation of elemental composition of the sample materials with relatively higher contrast ratio, whereas the ionization probe generated by picosecond laser pulses is more sensitive to the material density or structural change. These observations can be well explained by the different roles of multi-photon ionization and avalanche ionization involved in material breakdown.

Two-dimensional image sharpening in optical coherence tomography by deconvolution

Point spread functions (PSF) of a given OCT system in both longitudinal and transversal directions often cause bluring in OCT images. A two-dimensional sharpening method is proposed to simultaneously deconvolve both the longitudinal and transversal PSFs. This is achieved through employing an artificial kernel formed by a two-dimensional matrix, incoroporating the product of the longitudinal and transversal PSFs. For comparison, both Wiener and Lucy-Richardson algorithms are used for the two dimensional deconvolution. The experimental results show that these two deconvolution algorithms can enhance the image sharpness effectively in both longitudinal and transversal directions.

Application of fuzzy relations to color image retrieval

We report for the first time an application of fuzzy relations in fuzzy set theory to the color image retrieval. In the process, a new method of spatial partition of color space is developed for the comparisons. Our theoretical and experimental results show clearly the advantages of our approach in both accuracy and speed of the retrieval.

Fuzzy color image retrieval based on correlation

In this paper, we report a novel method of fuzzy color-image retrieval based on color histogram correlation, for which peaks of the histograms are first ranked according to their heights and then the correlation based on both height and position will be calculated for correspondent rank respectively. A correlation result will be defined by ?- cut procedures. Experiment results confirmed the effectiveness of the method.