A new CMOS active pixel image sensor is reported. The sensor uses a 2.0 p n double-poly, double-metal foundry CMOS process and is realized as a 128 x 128 array of 40 p n x 40 pm pixels. The sensor features TTL compatible voltages, low noise and large dynamic range, and will be useful in machine vision and smart sensor applications.

Pixel unmixing is commonly performed by employing a least squared (LS) error criterion, making it sensitive to outliers. As an alternative, the least median of squares (LMedS) method is proposed. Not only is it extremely robust, but it is efficient and straightforward both to implement and use.

1. Introduction CMOS image sensors have received much attention over the last decade, offering the promise of ultra low power and camera-on-chip integration in standard fabrication processes. This year over 30 million such sensors have been shipped with estimated annual growth rate of over 40%. Most CMOS image sensors shipped today are application-specific (e.g., optical mouse) or used in low-e...

In this paper, we propose a novel pixel-wise visual object tracking framework that can track any anonymous object in a noisy background. The framework consists of two submodels, a global attention model and a local segmentation model. The global model generates a region of interests (ROI) that the object may lie in the new frame based on the past object segmentation maps; while the local model ...

We present an image-conditional image generation model. The model transfers an input domain to a target domain in semantic level, and generates the target image in pixel level. To generate realistic target images, we employ the real/fake-discriminator as in Generative Adversarial Nets [6], but also introduce a novel domain-discriminator to make the generated image relevant to the input image. W...

We present a novel architecture for flexible control of shading rates in a GPU pipeline, and demonstrate substantially reduced shading costs for various applications. We decouple shading and visibility by restricting and quantizing shading rates to a finite set of screen-aligned grids, leading to simpler and fewer changes to the GPU pipeline compared to alternative approaches. Our architecture ...

Canonical minimal-pixel hologram geometries are presented in the context of a generalized image relay system. We illustrate that the number of pixels ultimately depends only on the physical parameters of the displayed image (the Smith-Lagrange optical invariant). Various choices of synthetic optical geometries lead to pixel aspect ratios that are found to be good matches to available display te...

In our affluent post-capitalist society in which basic existential needs are satisfied with plenty to spare and the workaday world is seen as sobering and disi l lusioning,1 people have a growing demand for emotional stimulation—in concrete terms, the value of money earned for services rendered is exchanged for emotional consumption experiences. Thus, the new type of “experience consumer” is no...

In this paper, we introduce Pixel Cache Light Tracing, which is a new low-noise combination of eye-path and light-path tracing. In the first pass, eye-path vertices are distributed from the observer and stored in a hit point map analogous to progressive photon mapping. In the second pass, photons are traced from the light source and projected to the image as well as gathered by the hit point ma...

We have developed several improved techniques because standard reduction software was inadequate for the Treasury Project on η Carinae. Though adapted specifically to HST/STIS/CCD data, some of these methods can be useful for images obtained with other instruments. For instance, our subpixel modeling technique reduces some of the evil effects of marginal sampling, i.e., in images where the pixe...