On-line learning of mutually orthogonal subspaces for face recognition by image sets.

We address the problem of face recognition by matching image sets. Each set of face images is represented by a subspace (or linear manifold) and recognition is carried out by subspace-to-subspace matching. In this paper, 1) a new discriminative method that maximises orthogonality between subspaces is proposed. The method improves the discrimination power of the subspace angle based face recognition method by maximizing the angles between different classes. 2) We propose a method for on-line updating the discriminative subspaces as a mechanism for continuously improving recognition accuracy. 3) A further enhancement called locally orthogonal subspace method is presented to maximise the orthogonality between competing classes. Experiments using 700 face image sets have shown that the proposed method outperforms relevant prior art and effectively boosts its accuracy by online learning. It is shown that the method for online learning delivers the same solution as the batch computation at far lower computational cost and the locally orthogonal method exhibits improved accuracy. We also demonstrate the merit of the proposed face recognition method on portal scenarios of multiple biometric grand challenge.

Liquid crystal over silicon device characteristics for holographic projection of high-definition television images.

We discuss some fundamental characteristics of a phase-modulating device suitable to holographically project a monochrome video frame with 1280 x 720 resolution. The phase-modulating device is expected to be a liquid crystal over silicon chip with silicon area similar to that of commercial devices. Its basic characteristics, such as number of pixels, bits per pixel, and pixel dimensions, are optimized in terms of image quality and optical efficiency. Estimates of the image quality are made from the noise levels and contrast, while efficiency is calculated by considering the beam apodization, device dead space, diffraction losses, and the sinc envelope.

Autonomous extraction of optimal flame fronts in OH planar laser-induced fluorescence images.

The location of a flame front is often taken as the point of maximum OH gradient. Planar laser-induced fluorescence of OH can be used to obtain the flame front by extracting the points of maximum gradient. This operation is typically performed using an edge detection algorithm. The choice of operating parameters a priori poses significant problems of robustness when handling images with a range of signal-to-noise ratios. A statistical method of parameter selection originating in the image processing literature is detailed, and its merit for this application is demonstrated. A reduced search space method is proposed to decrease computational cost and render the technique viable for large data sets. This gives nearly identical output to the full method. These methods demonstrate substantial decreases in data rejection compared to the use of a priori parameters. These methods are viable for any application where maximum gradient contours must be accurately extracted from images of species or temperature, even at very low signal-to-noise ratios.

A survey of deep-water coral and sponge habitats along the west coast of the US using a remotely operated vehicle: NOAA Fisheries Survey Vessel (FSV)'Bell M. Shimada', November 1-5, 2010

Remotely operated vehicle (ROV) surveys were conducted from NOAA’s state-of-the-art Fisheries Survey Vessel (FSV) Bell M. Shimada during a six-day transit November 1-5, 2010 between San Diego, CA and Seattle, WA. The objective of this survey was to locate and characterize deep-sea coral and sponge ecosystems at several recommended sites in support of NOAA’s Coral Reef Conservation Program. Deep-sea corals and sponges were photographed and collected whenever possible using the Southwest Fisheries Science Center’s (SWFSC) Phantom ROV ‘Sebastes’ (Fig. 1). The surveyed sites were recommended by National Marine Sanctuary (NMS) scientists at Monterey Bay NMS, Gulf of the Farallones NMS, and Olympic Coast NMS (Fig. 2). The specific sites were: Sur Canyon, The Football, Coquille Bank, and Olympic Coast NMS. During each dive, the ROV collected digital still images, video, navigation, and along-track conductivity-temperature-depth (CTD), and optode data. Video and high-resolution photographs were used to quantify abundance of corals, sponges, and associated fishes and invertebrates to the lowest practicable taxonomic level, and also to classify the seabed by substrate type. A reference laser system was used to quantify area searched and estimate the density of benthic fauna.