Modern digital cameras employ silicon focal plane array (FPA) image sensors featuring millions of pixels. However, it is possible to make a camera that only needs one pixel. In these cameras a spatial light modulator, placed before or after the object to be imaged, applies a time-varying pattern and synchronized intensity measurements are made with a single-pixel detector. The principle of compressed sensing then allows an image to be generated. As the approach suits a wide a variety of detector technologies, images can be collected at wavelengths outside the reach of FPA technology or at high frame rates or in three dimensions. Promising applications include the visualization of hazardous gas leaks and 3D situation awareness for autonomous vehicles. Rather than requiring millions of pixels, it is possible to make a camera that only needs one pixel. This Review details the working principle, advantages, technical considerations and future potential of single-pixel imaging.

Principles and prospects for single-pixel imaging

Microscopy is an essential tool in a huge range of research areas Until now, microscopy has been largely restricted to imaging in the visible region of the electromagnetic spectrum Here we present a microscope system that uses single-pixel imaging techniques to produce images simultaneously in the visible and shortwave infrared We apply our microscope to the inspection of various objects, including a silicon CMOS sensor, highlighting the complementarity of the visible and shortwave infrared wavebands The system is capable of producing images with resolutions between 32×32 and 128×128 pixels at corresponding frame rates between 10 and 06 Hz We introduce a compressive technique that does not require postprocessing, resulting in a predicted frame rate increase by a factor 8 from a compressive ratio of 125% with only 28% relative error

Single-pixel infrared and visible microscope

The use of convex optimization for the recovery of sparse signals from incomplete or compressed data is now common practice. Motivated by the success of basis pursuit in recovering sparse vectors, new formulations have been proposed that take advantage of different types of sparsity. In this paper we propose an efficient algorithm for solving a general class of sparsifying formulations. For several common types of sparsity we provide applications, along with details on how to apply the algorithm, and experimental results.

http://www.optimization-online.org/DB_FILE///2010/02/2537.pdf

Sparse Optimization with Least-Squares Constraints

Conventional cameras rely upon a pixelated sensor to provide spatial resolution. An alternative approach replaces the sensor with a pixelated transmission mask encoded with a series of binary patterns. Combining knowledge of the series of patterns and the associated filtered intensities, measured by single-pixel detectors, allows an image to be deduced through data inversion. In this work we extend the concept of a ‘single-pixel camera’ to provide continuous real-time video at 10 Hz , simultaneously in the visible and short-wave infrared, using an efficient computer algorithm. We demonstrate our camera for imaging through smoke, through a tinted screen, whilst performing compressive sampling and recovering high-resolution detail by arbitrarily controlling the pixel-binning of the masks. We anticipate real-time single-pixel video cameras to have considerable importance where pixelated sensors are limited, allowing for low-cost, non-visible imaging systems in applications such as night-vision, gas sensing and medical diagnostics.

/pdf/simultaneous-real-time-visible-and-infrared-video-with-3ro21t5woz.pdf

Simultaneous real-time visible and infrared video with single-pixel detectors

We present a compressive sensing video acquisition scheme that relies on the sparsity properties of video in the spatial domain. In this scheme, the video sequence is represented by a reference frame, followed by the difference of measurement results between each pair of neighboring frames. The video signal is reconstructed by first reconstructing the frame differences using 1 minimization algorithm, then adding them sequentially to the reference frame. Simulation results on both simulated and real video sequences show that when the spatial changes between neighboring frames are small, this scheme provides better reconstruction results than existing compressive sensing video acquisition schemes, such as 2-D or 3-D wavelet methods and the minimum total-variance (TV) method. This scheme is suitable for compressive sensing acquisition of video sequences with relatively small spatial changes. A method that estimates the amount of spatial change based on the statistical properties of measurement results is also presented.

Video compressive sensing using spatial domain sparsity

Compressive Sensing (CS) is a recently emerged signal processing method. It shows that when a signal is sparse in a certain basis, it can be recovered from a small number of random measurements made on it. In this work, we investigate the possibility of utilizing CS to sample the video stream acquired by a fixed surveillance camera in order to reduce the amount of data transmitted. For every 15 continuous video frames, we select the first frame in the video stream as the reference frame. Then for each following frame, we compute the difference between this frame and its preceding frame, resulting in a difference frame, which can be represented by a small number of measurement samples. By only transmitting these samples, we greatly reduce the amount of transmitted data. The original video stream can still be effectively recovered. In our simulations, SPGL1 method is used to recover the original frame. Two different methods, random measurement and 2D Fourier transform, are used to make the measurements. In our simulations, the Peak Signal-to-Noise Ratio (PSNR) ranges from 28.0dB to 50.9dB, depending on the measurement method and number of measurement used, indicating good recovery quality. Besides a good compression rate, the CS technique has the properties of being robust to noise and easily encrypted which all make CS technique a good candidate for signal processing in communication.

The application of Compressive Sensing technique on a stationary surveillance camera system

Compressive Sensing (CS) theory shows that if a signal is sparse under a certain basis, it can be recovered from a small number of measurement samples. These measurements can be significant departures from the traditional notion of a pixel associated with a single detector element in a focal plane array. In this work, we study the problem of how different sampling methods affect image recovery. In our study, a 2D Fourier transform is used as the measurement method. Spatial frequency sampling is accomplished using three different methods, uniform random method, 2D Gaussian method with different variances and a LineMask method. Infrared images from a stationary surveillance camera are recovered from these collected samples. Peak Signal-to-Noise Ratio (PSNR) is used to evaluate the quality of recovered images. Our simulation results show that, with the same number of collected measurement samples, both LineMask and 2D Gaussian methods offer better image recovery results than the random method. For the 2D Gaussian method, the image recovery results improve slightly as the variance of Gaussian sampling function decreases. The recovery result of the LineMask method is between the best and worst cases from the Gaussian method. Our results show that while the CS technique allows images to be recovered from randomly chosen measurement samples, the method used to collect the measurement samples does affect the signal recovery quality. Choosing a proper sampling method can optimize recovery using the CS technique.

Jing Zheng

Papers

Video compressive sensing using spatial domain sparsity

The application of Compressive Sensing technique on a stationary surveillance camera system

Application of compressive sensing theory in infrared imaging systems