Peter Wai Ming Tsang

Holographic video at 40 frames per second for 4-million object points.

We propose a fast method for generating digital Fresnel holograms based on an interpolated wavefront-recording plane (IWRP) approach. Our method can be divided into two stages. First, a small, virtual IWRP is derived in a computational-free manner. Second, the IWRP is expanded into a Fresnel hologram with a pair of fast Fourier transform processes, which are realized with the graphic processing unit (GPU). We demonstrate state-of-the-art experimental results, capable of generating a 2048 x 2048 Fresnel hologram of around 4 × 10(6) object points at a rate of over 40 frames per second.

Complex Fresnel hologram display using a single SLM

We propose a novel optical method to display a complex Fresnel hologram using a single spatial light modulator (SLM). The method consists of a standard coherent image processing system with a sinusoidal grating at the Fourier plane. Two or three position-shifted amplitude holograms displayed at the input plane of the processing system can be coupled via the grating and will be precisely overlapped at the systems output plane. As a result, we can synthesize a complex hologram that is free of the twin image and the zero-order light using a single SLM. Because the twin image is not removed via filtering, the full bandwidth of the SLM can be utilized for displaying on-axis holograms. In addition, the degree of freedom of the synthesized complex hologram display can be extended by involving more than three amplitude holograms.

Novel method for converting digital Fresnel hologram to phase-only hologram based on bidirectional error diffusion

We report a novel and fast method for converting a digital, complex Fresnel hologram into a phase-only hologram. Briefly, the pixels in the complex hologram are scanned sequentially in a row by row manner. The odd and even rows are scanned from opposite directions, constituting to a bidirectional error diffusion process. The magnitude of each visited pixel is forced to be a constant value, while preserving the exact phase value. The resulting error is diffused to the neighboring pixels that have not been visited before. The resulting novel phase-only hologram is called the bidirectional error diffusion (BERD) hologram. The reconstructed image from the BERD hologram exhibits high fidelity as compared with those obtained with the original complex hologram.

A genetic algorithm for aligning object shapes

Recently, the use of dominant points for boundary alignment has been widely adopted in a lot of object recognition techniques. The success of these approaches is highly dependent on the availability of a set of spatially matched dominant point pairs on the scene and the reference contours. This criteria, however, is difficult to attain in practice as the distribution of dominant points are often found to change with the pose and size of the object images that are grabbed under different camera position. In this paper, a novel technique based on the genetic algorithm for searching the best alignment between contours of near-planar objects is reported. The method is more efficient and robust than the dominant point approaches, and is capable of arriving at the optimal solution instead of being trapped in the local minimum where only partial alignment of the contours is achieved. Experimental results obtained with the proposed scheme are encouraging which demonstrate the feasibility of the approach.

Fast conversion of digital Fresnel hologram to phase-only hologram based on localized error diffusion and redistribution

Past research has demonstrated that a digital, complex Fresnel hologram can be converted into a phase-only hologram with the use of the bi-direction error diffusion (BERD) algorithm. However, the recursive nature error diffusion process is lengthy and increases monotonically with hologram size. In this paper, we propose a method to overcome this problem. Briefly, each row of a hologram is partitioned into short non-overlapping segments, and a localized error diffusion algorithm is applied to convert the pixels in each segment into phase only values. Subsequently, the error signal is redistributed with low-pass filtering. As the operation on each segment is independent of others, the conversion process can be conducted at high speed with the graphic processing unit. The hologram obtained with the proposed method, known as the Localized Error Diffusion and Redistribution (LERDR) hologram, is over two orders of magnitude faster than that obtained by BERD for a 2048×2048 hologram, exceeding the capability of generating quality phase-only holograms in video rate.

eXploratory K-Means: A new simple and efficient algorithm for gene clustering

In this paper, a novel gene expression clustering method known as eXploratory K-Means (XK-Means) is proposed. The method is based on the integration of the K-Means framework, and an exploratory mechanism to prevent premature convergence of the clustering process. Experimental results reveal that the performance of XK-Means in grouping gene expressions, measured in terms of speed, error and stability, is superior to existing methods that are based on evolutionary algorithm. In addition, the complexity of the proposed method is lower and the method can be easily implemented in practice.

A genetic algorithm for affine invariant recognition of object shapes from broken boundaries

Abstract This paper presents a novel technique for affine invariant recognition of single, near planar object shapes from broken boundaries. In this approach, the matching scores between pairs of object contours are computed based on a combination of a genetic algorithm and a distance transform.

A genetic algorithm for projective invariant object recognition

A genetic algorithm is developed for recognition of near planar objects under projective transformation. The approach can be generalised in any 2D geometric transformation. No approximation to perspective projection is required, so the camera used for capturing the objects is allowed to locate at arbitrary position away from the scene. A genetic algorithm is employed to search for the optimal solution in order to identify the object and to estimate its location with reference to the model. Encouraging results were obtained by testing the approach from a number of industrial handtools.

Generation of phase-only Fresnel hologram based on down-sampling

We present a novel non-iterative method for generating phase-only Fresnel holograms. The intensity image of the source object scene is first down-sampled with uniform grid-cross lattices. A Fresnel hologram is then generated from the intensity and the depth information of the sampled object points. Subsequently, only the phase component of the hologram is preserved, resulting in a pure phase hologram that we call the sampled-phase-only hologram (SPOH). Experimental evaluation reveals that the numerical, as well as the optical reconstructed images of the proposed phase-only hologram derived with our method are of high visual quality. Moreover, the reconstructed optical image is brighter, and less affected by phase noise contamination on the hologram as compared with those generated with existing error-diffusion approaches.

Fast generation of Fresnel holograms based on multirate filtering

One of the major problems in computer-generated holography is the high computation cost involved for the calculation of fringe patterns. Recently, the problem has been addressed by imposing a horizontal parallax only constraint whereby the process can be simplified to the computation of one-dimensional sublines, each representing a scan plane of the object scene. Subsequently the sublines can be expanded to a two-dimensional hologram through multiplication with a reference signal. Furthermore, economical hardware is available with which sublines can be generated in a computationally free manner with high throughput of approximately 100 M pixels/second. Apart from decreasing the computation loading, the sublines can be treated as intermediate data that can be compressed by simply downsampling the number of sublines. Despite these favorable features, the method is suitable only for the generation of white light (rainbow) holograms, and the resolution of the reconstructed image is inferior to the classical Fresnel hologram. We propose to generate holograms from one-dimensional sublines so that the above-mentioned problems can be alleviated. However, such an approach also leads to a substantial increase in computation loading. To overcome this problem we encapsulated the conversion of sublines to holograms as a multirate filtering process and implemented the latter by use of a fast Fourier transform. Evaluation reveals that, for holograms of moderate size, our method is capable of operating 40,000 times faster than the calculation of Fresnel holograms based on the precomputed table lookup method. Although there is no relative vertical parallax between object points at different distance planes, a global vertical parallax is preserved for the object scene as a whole and the reconstructed image can be observed easily.