Method And Apparatus For Compressive Imaging Device Having Startle Reflex

Abstract

A method and apparatus for new digital image / video camera that directly acquires random projections of the incident light field without first collecting the pixels / voxels and has a startle reflex. The startle reflex is achieved by collecting a training set of samples while operating the imaging device at the low sampling rate, computing an empirical probability distribution of the training set of samples, collecting a testing set of samples while operating the imaging device at the low sampling rate, computing an empirical probability distribution of the testing set of samples, computing an empirical entropy test statistic, and capturing an image at the high sampling rate in response to the computed empirical entropy statistic exceeding a threshold.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS[0001]The present application is a continuation of U.S. Nonprovisional patent application Ser. No. 11 / 379,688, filed on Apr. 21, 2006, which claimed the benefit of the filing dates of U.S. Provisional Application Ser. No. 60 / 673,364 entitled “Method and Apparatus for Optical Image Compression,” and filed on Apr. 21, 2005; U.S. Provisional Application Ser. No. 60 / 679,237 entitled “Method and Apparatus for Reconstructing Data from Multiple Sources,” and filed on May 10, 2005; U.S. Provisional Application Ser. No. 60 / 729,983 entitled “Random Filters for Compressive Sampling and Reconstruction,” and filed on Oct. 25, 2005; U.S. Provisional. Application Ser. No. 60 / 732,374 entitled “Method and Apparatus for Compressive Sensing for Analog-to-Information Conversion,” and filed on Nov. 1, 2005; U.S. Provisional Application Ser. No. 60 / 735,616 entitled “Method and Apparatus for Distributed Compressed Sensing,” and filed on Nov. 10, 2005; and U.S. Provis...

AI Technical Summary

Benefits of technology

[0014]A small number of detectors, even a single detector, can be used. Thus, the camera can be adapted to image at wavelengths of electromagnetic radiation that are currently impossible with conventional CCD and CMOS imagers. This feature is particularly advantageous, because in some cases the usage of many detectors is impossible or impractical, whereas the usage of a small number of detectors, or even a single detector, may become feasible using compressive imaging.

[0015]A camera in accordance with the present invention can also be used to take streaming measurements of a video signal, which can then be recovered using CS techniques designed for either 2-dimensional (2D) frame-by-frame reconstruction or joint 3D reconstruction. This allows a significant reduction in the computational complexity of the video encoding process.

[0017]Potentially single detector or small number of detectors: By time-multiplexing each detector, we can use a less expensive and yet more sensitive photon detectors. This is particularly important when the detector is expensive, making an N-pixel array prohibitive. A single detector camera can also be adapted to image at wavelengths that are currently impossible with conventional CCD and CMOS imagers.

[0020]Robustness and progressivity: Random and pseudorandom measurements are robust in the sense that the measurements have equal priority, unlike the Fourier or wavelet coefficients in current transform coders. Thus they allow a progressively better reconstruction of the data as more measurements are obtained; one or more measurements can also be lost without corrupting the entire reconstruction.

Problems solved by technology

First, acquiring large amounts of raw image or video data (large N) can be expensive, particularly at wavelengths where CMOS or CCD sensing technology is limited.

Second, compressing raw data can be computationally demanding, particularly in the case of video.

In the former case, scheduled power-off periods could result in missing an important event entirely.

In the latter case, we require additional hardware that may be costly or undesirable.

Moreover, in both cases the system suffers from a “power-on lag,” which delays image or video capture, potentially causing the camera to miss the important event.

This kind of scheme is impossible in the traditional digital camera paradigm, which is an all-or-nothing scheme: either an image / video is captured at full rate, or no image / video is captured at all.

Images