On sparsity and overcompleteness in image models

Computational models of visual cortex, and in particular those based on sparse coding, have enjoyed much recent attention. Despite this currency, the question of how sparse or how over-complete a sparse representation should be, has gone without principled answer. Here, we use Bayesian model-selection methods to address these questions for a sparse-coding model based on a Student-t prior. Having validated our methods on toy data, we find that natural images are indeed best modelled by extremely sparse distributions; although for the Student-t prior, the associated optimal basis size is only modestly over-complete.

Beyond 2D-grids: A dependence maximization view on image browsing

Ideally, one would like to perform image search using an intuitive and friendly approach. Many existing image search engines, however, present users with sets of images arranged in some default order on the screen, typically the relevance to a query, only. While this certainly has its advantages, arguably, a more flexible and intuitive way would be to sort images into arbitrary structures such as grids, hierarchies, or spheres so that images that are visually or semantically alike are placed together. This paper focuses on designing such a navigation system for image browsers. This is a challenging task because arbitrary layout structure makes it difficult - if not impossible - to compute cross-similarities between images and structure coordinates, the main ingredient of traditional layouting approaches. For this reason, we resort to a recently developed machine learning technique: kernelized sorting. It is a general technique for matching pairs of objects from different domains without requiring cross-domain similarity measures and hence elegantly allows sorting images into arbitrary structures. Moreover, we extend it so that some images can be preselected for instance forming the tip of the hierarchy allowing to subsequently navigate through the search results in the lower levels in an intuitive way. Copyright 2010 ACM.

Time-resolved particle image velocimetry within the nozzle of a drop-on-demand printhead

Time-resolved particle image velocimetry (PIV) has been performed inside the nozzle of a commercially available inkjet print-head to obtain the time-dependent velocity waveform. A printhead with a single transparent nozzle 80 μm in orifice diameter was used to eject single droplets at a speed of 5 m/s. An optical microscope was used with an ultra-high-speed camera to capture the motion of particles suspended in a transparent liquid at the center of the nozzle and above the fluid meniscus at a rate of half a million frames per second. Time-resolved velocity fields were obtained from a fluid layer approximately 200 μm thick within the nozzle for a complete jetting cycle. A Lagrangian finite-element numerical model with experimental measurements as inputs was used to predict the meniscus movement. The model predictions showed good agreement with the experimental results. This work provides the first experimental verification of physical models and numerical simulations of flows within a drop-on-demand nozzle. © 2012 Society for Imaging Science and Technology.

How to achieve ultra high photoconductive gain for transparent oxide semiconductor image sensors

We present quantitative analysis of the ultra-high photoconductivity in amorphous oxide semiconductor (AOS) thin film transistors (TFTs), taking into account the sub-gap optical absorption in oxygen deficiency defects. We analyze the basis of photoconductivity in AOSs, explained in terms of the extended electron lifetime due to retarded recombination as a result of hole localization. Also, photoconductive gain in AOS photo-TFTs can be maximized by reducing the transit time associated with short channel lengths, making device scaling favourable for high sensitivity operation. © 2012 IEEE.

Rapid hologram generation utilizing layer-based approach and graphic rendering for realistic three-dimensional image reconstruction by angular tiling

An approach of rapid hologram generation for the realistic three-dimensional (3-D) image reconstruction based on the angular tiling concept is proposed, using a new graphic rendering approach integrated with a previously developed layer-based method for hologram calculation. A 3-D object is simplified as layered cross-sectional images perpendicular to a chosen viewing direction, and our graphics rendering approach allows the incorporation of clear depth cues, occlusion, and shading in the generated holograms for angular tiling. The combination of these techniques together with parallel computing reduces the computation time of a single-view hologram for a 3-D image of extended graphics array resolution to 176 ms using a single consumer graphics processing unit card. © 2014 SPIE and IS and T.

Rapid patterning of 1-D collagenous topography as an ECM protein fibril platform for image cytometry.

Cellular behavior is strongly influenced by the architecture and pattern of its interfacing extracellular matrix (ECM). For an artificial culture system which could eventually benefit the translation of scientific findings into therapeutic development, the system should capture the key characteristics of a physiological microenvironment. At the same time, it should also enable standardized, high throughput data acquisition. Since an ECM is composed of different fibrous proteins, studying cellular interaction with individual fibrils will be of physiological relevance. In this study, we employ near-field electrospinning to create ordered patterns of collagenous fibrils of gelatin, based on an acetic acid and ethyl acetate aqueous co-solvent system. Tunable conformations of micro-fibrils were directly deposited onto soft polymeric substrates in a single step. We observe that global topographical features of straight lines, beads-on-strings, and curls are dictated by solution conductivity; whereas the finer details such as the fiber cross-sectional profile are tuned by solution viscosity. Using these fibril constructs as cellular assays, we study EA.hy926 endothelial cells' response to ROCK inhibition, because of ROCK's key role in the regulation of cell shape. The fibril array was shown to modulate the cellular morphology towards a pre-capillary cord-like phenotype, which was otherwise not observed on a flat 2-D substrate. Further facilitated by quantitative analysis of morphological parameters, the fibril platform also provides better dissection in the cells' response to a H1152 ROCK inhibitor. In conclusion, the near-field electrospun fibril constructs provide a more physiologically-relevant platform compared to a featureless 2-D surface, and simultaneously permit statistical single-cell image cytometry using conventional microscopy systems. The patterning approach described here is also expected to form the basics for depositing other protein fibrils, seen among potential applications as culture platforms for drug screening.

A 1,000 Frames/s Programmable Vision Chip with Variable Resolution and Row-Pixel-Mixed Parallel Image Processors

A programmable vision chip with variable resolution and row-pixel-mixed parallel image processors is presented. The chip consists of a CMOS sensor array, with row-parallel 6-bit Algorithmic ADCs, row-parallel gray-scale image processors, pixel-parallel SIMD Processing Element (PE) array, and instruction controller. The resolution of the image in the chip is variable: high resolution for a focused area and low resolution for general view. It implements gray-scale and binary mathematical morphology algorithms in series to carry out low-level and mid-level image processing and sends out features of the image for various applications. It can perform image processing at over 1,000 frames/s (fps). A prototype chip with 64 x 64 pixels resolution and 6-bit gray-scale image is fabricated in 0.18 mu m Standard CMOS process. The area size of chip is 1.5 mm x 3.5 mm. Each pixel size is 9.5 mu m x 9.5 mu m and each processing element size is 23 mu m x 29 mu m. The experiment results demonstrate that the chip can perform low-level and mid-level image processing and it can be applied in the real-time vision applications, such as high speed target tracking.

Blind image restoration based on the theory of high-dimensional space geometry

In practical situations, the causes of image blurring are often undiscovered or difficult to get known. However, traditional methods usually assume the knowledge of the blur has been known prior to the restoring process, which are not practicable for blind image restoration. A new method proposed in this paper aims exactly at blind image restoration. The restoration process is transformed into a problem of point distribution analysis in high-dimensional space. Experiments have proved that the restoration could be achieved using this method without re-knowledge of the image blur. In addition, the algorithm guarantees to be convergent and has simple computation.