Fig 8 - uploaded by Changyin Zhou
Content may be subject to copyright.
![Light field camera using an array of pairs of negative lenses and prisms [52]. (a) Geometry of the light field camera. (b) Prototype of the lens-prism array. (c) From left to right are the intermediate light fields before the lens-prism array, after the array, and after the aperture, respectively. Similar to Fig. 7, the prism arrays will split the input light field and create multiple copies; each of them capturing one low-dimensional slice of the high-dimensional light field.](profile/Changyin-Zhou/publication/224263848/figure/fig11/AS:668678165110788@1536436706535/Light-field-camera-using-an-array-of-pairs-of-negative-lenses-and-prisms-52-a.jpg)
Light field camera using an array of pairs of negative lenses and prisms [52]. (a) Geometry of the light field camera. (b) Prototype of the lens-prism array. (c) From left to right are the intermediate light fields before the lens-prism array, after the array, and after the aperture, respectively. Similar to Fig. 7, the prism arrays will split the input light field and create multiple copies; each of them capturing one low-dimensional slice of the high-dimensional light field.
Source publication
A computational camera uses a combination of optics and processing to produce images that cannot be captured with traditional cameras. In the last decade, computational imaging has emerged as a vibrant field of research. A wide variety of computational cameras has been demonstrated to encode more useful visual information in the captured images, as...
Contexts in source publication
Context 1
... Light Field Acquisition: Georgeiv et al. [52] propose using an array of lens-prism pairs in front of the main lens to capture light fields, as shown in Fig. 8. The geometry of the camera is shown in Fig. 8(a). Each prism in the array has a different angle of deviation; therefore, similar to the biprism design (see Fig. 7), the prism array splits the FOV into multiple pieces, i.e., each corresponding to a different viewpoint. In other words, the camera observes the same small FOV but from ...
Context 2
... Light Field Acquisition: Georgeiv et al. [52] propose using an array of lens-prism pairs in front of the main lens to capture light fields, as shown in Fig. 8. The geometry of the camera is shown in Fig. 8(a). Each prism in the array has a different angle of deviation; therefore, similar to the biprism design (see Fig. 7), the prism array splits the FOV into multiple pieces, i.e., each corresponding to a different viewpoint. In other words, the camera observes the same small FOV but from different viewpoints. The information captured by the ...
Similar publications
In an earlier publication [Appl. Opt. 47, 3722 (2008)] we suggested an adaptive optical lens, which consists of two cascaded diffractive optical elements (DOEs). Due to the Moiré-effect the combined optical element acts as a Fresnel zone lens with a refractive power that can be continuously adjusted by a mutual rotation of the two stacked DOEs. Her...
Regulating the intensity distribution of an extended source to produce a prescribed illumination in three-dimensional (3D) rotationally symmetric geometry remains a challenging issue in illumination design. In this Letter, we present an effective method focusing on creating prescribed intensity designs for extended sources. By this method, a prescr...
The MAP-Elites algorithm produces a set of high-performing solutions that vary according to features defined by the user. This technique to 'illuminate' the problem space through the lens of chosen features has the potential to be a powerful tool for exploring design spaces, but is limited by the need for numerous evaluations. The Surrogate-Assiste...
The paper describes the design and realization of optical systems for concentrating solar light onto an optical fibre. Three different classes of optical system, with increasing surface complexity, were studied. These were collectors with spherical surfaces only, parabolic concentrators and collectors containing aspherical surfaces. This research l...
Citations
... A wide selection of snapshot image formation methods are available in the literature [9], [10], which demands adequate techniques to process the acquisitions in a prompt and flexible manner. As the manufacture of many of such prototypes can be seen as a combination of elemental components, this work proposes to describe them under a unified mathematical reconstruction framework. ...
Novel optical imaging devices allow for hybrid acquisition modalities such as compressed acquisitions with locally different spatial and spectral resolutions captured by the same focal plane array. In this work, we propose to model a multiresolution coded acquisition (MRCA) in a generic framework, which natively includes acquisitions by conventional systems such as those based on spectral/color filter arrays, compressed coded apertures, and multiresolution sensing. We propose a model-based image reconstruction algorithm performing a joint demosaicing and fusion (JoDeFu) of any acquisition modeled in the MRCA framework. The JoDeFu reconstruction algorithm solves an inverse problem with a proximal splitting technique and is able to reconstruct an uncompressed image datacube at the highest available spatial and spectral resolution. An implementation of the code is available at
https://github.com/danaroth83/jodefu
.
... For example, the camera lens is first improved to obtain the amount of defocus blur [16], and then the deconvolution algorithm is used to obtain the depth information of the scene. A Gaussian point spread function model with known defocus blur parameters was used to re-blur a defocus blur image twice, two images are obtained with different defocus blur levels [17], and then the ratio of the difference is calculated between the defocus blur of the two images and the original defocus blur image, the maximum value of the ratio is used to obtain the amount of defocus blur at the edge of the image, the panoramic depth information of the target scene is estimated finally. ...
Most of the depth restoration algorithms are complicated for a single defocused image, the restoration effect is poor on the image edges, complex textures and shadow areas. In this paper, the depth of a scene is recovered by a single image defocus cue, a novel approach of single defocused image depth restoration is proposed based on superpixel segmentation. First, the simple linear iterative cluster (SLIC) is used to divide the original image into several superpixel modules. Then, the defocus blur amount of each superpixel module is obtained according to the Gaussian–Cauchy mixed model, so as to obtain the superpixel-level sparse depth map. Finally, the cellular automata model is used to optimize the obtained sparse depth map and to restore the true and accurate panoramic depth map. Compared with other methods, the algorithm not only minimizes the error, but also simplifies the process of extending the edge defocus blur to the global. The experimental results on real data show that the method is not only less time-consuming, but also can effectively improve the depth recovery effect in areas with unclear edges, complex textures and shadows. These demonstrate the effectiveness of the method in providing reliable estimates of scene depth.
... In general, there are inherent losses of information when 3D objects are projected onto a 2D image plane such as the sensor in a digital camera (i.e., projective transformation) (Cao et al., 2011). This results from the sensor in a digital camera being restrictive in collecting the array of a complete set of light rays that occur in a 3D space (Zhou & Nayar, 2011). Accordingly, for this study, 2D vertical heights of icebergs that were measured were selected to allow the apex and the base of the iceberg to be in the same plane. ...
Long‐term monitoring programs to evaluate climate‐driven changes to tidewater glaciers, an important habitat for harbor seals (Phoca vitulina) in Alaska, are primarily carried out by costly, weather‐dependent aerial surveys from fixed‐winged aircraft. Unmanned aerial systems (UAS) can be an alternative cost‐effective solution for gathering image data to quantify, monitor, and manage these habitats. However, there is a paucity of information related to the accuracy of using imagery collected by UAS for purposes of measuring floating icebergs. We evaluated the accuracy of using a UAS with a built‐in 20‐megapixel (MP) camera as well as a consumer‐grade digital 16‐MP camera to capture images of floating and stationary icebergs for the purpose of collecting vertical height measurements. Images (n = 869) were captured of simulated icebergs (cuboidal foam boxes, Cb) (n = 5) and real icebergs (n = 5) that were either grounded or floating. The mean error ratios (Ers) obtained were less than 10% and derived by comparing the mean calculated measurements of heights of Cb obtained from images captured by UAS with the physically measured heights of these Cb. The mean Er for height measurements of grounded icebergs (n = 4) and one floating iceberg was also less than 10%. Within an object–image distance range of 6–25 m, the cameras captured images that were suitable to accurately calculate the heights of floating and grounded objects, and drift or uncontrolled movement of the UAS caused by wind or temporary loss of GPS did not have any effect on measurement error. Our study provides substantial evidence of the accuracy associated with using images captured by UAS for measuring dimensions of structures positioned on either water or land surfaces. Ultimately, accurate surveys of glacial ice used by harbor seals will improve our understanding of the role of decreasing habitat in explaining population variability between different tidewater glaciers.
... However, these models are designed for multiple image DFD and are not suited for unconventional camera. Several authors have proposed performance models for unconventional camera, particularly for coded aperture [1][2][3]18 . However, these models only consider the aperture shape and not the remaining of the optical system or the processing parameters and they only compare the results of depth estimation on different forms of coded aperture without providing a real accuracy nor information about the depth range of their systems. ...
... Light field (LF) imaging [1] allows the representation of a wider variety of information when compared with conventional 2D cameras, by capturing both the scene radiance and the directionality of light rays reaching the camera sensor. The extra information captured by lenslet LF cameras and high density camera arrays (HDCA) [2] enables a plethora of postprocessing operations, such as the extraction of depth maps [3], image rendering with different focal planes, depth-of-field, and viewing perspectives [4,5,6]. ...
The adoption of light field (LF) technology is often hindered by the large amount of data that is necessary to represent such information. To alleviate such burden, specific coding techniques have been under research in recent years. However, most of them require decoding the whole LF data, even in applications where only a region of the LF visual content needs to be accessed. Therefore, encoding schemes that incorporate scalability and random access capabilities would cater to a wider range of LF use cases, particularly those requiring lossless compression. To enable such type of functionalities, an encoder must be designed so that the decoding process can extract only specific views/regions of the complete LF, according to user options or application requirements. This paper proposes hierarchical Minimum Rate Predictors (H-MRP), a lossless light field codec that provides the aforementioned functionalities. The main contributions herein proposed are: (i) the lossless encoding the LF in multiple hierarchical layers; (ii) the flexible selection of reference images for prediction; and (iii) the use disparity compensation between the reference images. The experimental results show that the proposed encoder provides a good compromise in terms of compression efficiency versus random access penalty, surpassing its best competitor by 25% in terms of random access and by 1 bit-per-pixel in terms of compression efficiency. When optimised to provide the best compression efficiency, the method surpasses current state-of-the-art lossless encoding by up to 3 bits-per-pixel.
... The two planes represented in Figure 1 are discretised allowing for only a finite number of viewpoints and light rays to be recorded. Several processes for capturing LFs exist, as described by Zhou in [26]. For the purposes of this paper, single sensor LF cameras and array of cameras are considered. ...
Common representations of light fields use four-dimensional data structures, where a given pixel is closely related not only to its spatial neighbours within the same view, but also to its angular neighbours, co-located in adjacent views. Such structure presents increased redundancy between pixels, when compared with regular single-view images. Then, these redundancies are exploited to obtain compressed representations of the light field, using prediction algorithms specifically tailored to estimate pixel values based on both spatial and angular references. This paper proposes new encoding schemes which take advantage of the four-dimensional light field data structures to improve the coding performance of Minimum Rate Predictors. The proposed methods expand previous research on lossless coding beyond the current state-of-the-art. The experimental results, obtained using both traditional datasets and others more challenging, show bit-rate savings no smaller than 10%, when compared with existing methods for lossless light field compression.
... Implementing deep learning (convolution) in free-space optics is reminiscent of classic computational optics and computational photography, which co-designs optics with image processing and vision algorithms [35]. Recent progresses in deep learning has pushed this trend even further to co-train optical parameters and DNN-based vision algorithms in an end-toend fashion. ...
... A typical setup for a free-space optical DNN is similar to computational cameras [35,45], where a custom-designed lens directly interrogates the real-world scene as an optical input signal. The lens executes the first convolution or fully connected layer of a DNN; the layer output is captured by an image sensor and sent to the digital system to execute the rest of the DNN model. ...
Deep learning using convolutional neural networks (CNNs) has been shown to significantly outperform many conventional vision algorithms. Despite efforts to increase the CNN efficiency both algorithmically and with specialized hardware, deep learning remains difficult to deploy in resource-constrained environments. In this paper, we propose an end-to-end framework to explore how to optically compute the CNNs in free-space, much like a computational camera. Compared to existing free-space optics-based approaches that are limited to processing single-channel (i.e., gray scale) inputs, we propose the first general approach, based on nanoscale metasurface optics, that can process RGB input data. Our system achieves up to an order of magnitude energy savings and simplifies the sensor design, all the while sacrificing little network accuracy.
... The two planes represented in Figure 1 are discretised allowing for only a finite number of viewpoints and light rays to be recorded. Several processes for capturing LF exist, as described by Zhou in [21]. For the purposes of this paper, single sensor LF cameras and array of cameras are considered. ...
Common representations of light fields use four-dimensional data structures, where a given pixel is closely related not only to its spatial neighbours within the same view, but also to its angular neighbours, co-located in adjacent views. Such structure presents increased redundancy between pixels, when compared with regular single-view images. Then, these redundancies are exploited to obtain compressed representations of the light field, using prediction algorithms specifically tailored to estimate pixel values based on both spatial and angular references. This paper proposes new encoding schemes which take advantage of the four-dimensional light field data structures to improve the coding performance of Minimum Rate Predictors. The proposed methods expand previous research on lossless coding beyond the current state-of-the-art. The experimental results, obtained using both traditional datasets and others more challenging, show bit-rate savings no smaller than 10%, when compared with existing methods for lossless light field compression.
... micro-lens-based and mask-based cameras. A taxonomy of computational light field cameras, according to the approach used to project the 4D light field onto a 2D sensor, can be found in [23]. ...
We present an all-in-one camera model that encompasses the architectures of most existing compressive-sensing light-field cameras, equipped with a single lens and multiple amplitude coded masks that can be placed at different positions between the lens and the sensor. The proposed model, named the equivalent multi-mask camera (EMMC) model, enables the comparison between different camera designs, e.g using monochrome or CFA-based sensors, single or multiple acquisitions, or varying pixel sizes, via a simple adaptation of the sampling operator. In particular, in the case of a camera equipped with a CFA-based sensor and a coded mask, this model allows us to jointly perform color demosaicing and light field spatio-angular reconstruction. In the case of variable pixel size, it allows to perform spatial super-resolution in addition to angular reconstruction. While the EMMC model is generic and can be used with any reconstruction algorithm, we validate the proposed model with a dictionary-based reconstruction algorithm and a regularization-based reconstruction algorithm using a 4D Total Variation-based regularizer for light field data. Experimental results with different reconstruction algorithms show that the proposed model can flexibly adapt to various sensing schemes. They also show the advantage of using an in-built CFA sensor with respect to monochrome sensors classically used in the literature.
... The two planes represented in Figure 1 are discretised allowing for only a finite number of viewpoints and light rays to be recorded. Several processes for capturing LF exist, as described by Zhou in [21]. For the purposes of this paper, single sensor S e n s o r p l a n e t s P 1,2 P 1,1 F o c u s p l a n e LF cameras and array of cameras are considered. ...
