Blur Compensation Research Articles

A Multi-aperture Coaxial Projector Balancing Shadow Suppression and Deblurring.

This paper proposes a projection system that optically removes the cast shadow in projection mapping. Specifically, we realize the large-aperture (LA) projection using a large-format Fresnel lens to suppress cast shadows by condensing the projection light from a wide viewing angle. However, the resolution and contrast of the projected results are significantly degraded by defocus blur, veiling glare, and stray light caused by the aberration of an LA Fresnel lens. To solve the technical problems, we employ two different approaches: optical and digital image processing methods. First, we introduce a residual projector with a typical aperture lens on the same optical axis as the LA projector, projecting the residual (i.e., high-frequency) components attenuated in the LA projection. These projectors play different roles in shadow suppression and blur compensation, both achieved by projecting simultaneously. Secondly, we optimize the pair of projection images that can balance the shadow suppression and deblurring performance of our projection system. We implemented a proof-of-concept prototype and validated the above-mentioned techniques through projection experiments and a user study.

Retrieving Object Motions From Coded Shutter Snapshot in Dark Environment.

Video object detection is a widely studied topic and has made significant progress in the past decades. However, the feature extraction and calculations in existing video object detectors demand decent imaging quality and avoidance of severe motion blur. Under extremely dark scenarios, due to limited sensor sensitivity, we have to trade off signal-to-noise ratio for motion blur compensation or vice versa, and thus suffer from performance deterioration. To address this issue, we propose to temporally multiplex a frame sequence into one snapshot and extract the cues characterizing object motion for trajectory retrieval. For effective encoding, we build a prototype for encoded capture by mounting a highly compatible programmable shutter. Correspondingly, in terms of decoding, we design an end-to-end deep network called detection from coded snapshot (DECENT) to retrieve sequential bounding boxes from the coded blurry measurements of dynamic scenes. For effective network learning, we generate quasi-real data by incorporating physically-driven noise into the temporally coded imaging model, which circumvents the unavailability of training data and with high generalization ability on real dark videos. The approach offers multiple advantages, including low bandwidth, low cost, compact setup, and high accuracy. The effectiveness of the proposed approach is experimentally validated under low illumination vision and provide a feasible way for night surveillance.

Angle of View Switching Method at High-Speed Using Motion Blur Compensation for Infrastructure Inspection

Efficient imaging is achieved under conditions of high relative velocity between the camera and the subject by using the following imaging system; two galvanometer mirrors are placed vertically in front of the camera, one for motion blur compensation and the other for switching the angle of view. The proposed system can overcome the shortcomings of conventional imaging systems with motion blur compensation, such as a small angle of view, and efficiently acquire high-resolution images. If the angle is changed for each capture while the mirrors are stationary for the exposure time, the natural frequency of the mirrors produces noise, leading to a poor resolution. However, this issue can be managed by generating and using an input that does not contain a natural frequency component. A target moving in one dimension can be captured and it is confirmed that the angle of view was extended from the obtained image. It is expected that the camera will be used for inspections under conditions where the relative speed between the camera and target is high, such as in highway tunnel inspections.

Small-scale infrared surround scan system with image motion blur compensation based on multisegment optical wedges

We study an infrared surround scan system based on a photodetector array with an entrance pupil of a small diameter with no specific requirements for achieving versatility and a high dynamic performance. A cooled midinfrared photodetector with an objective and a compensator for the tangential motion blur of an image are housed within a rotating unit. Command and data exchange are ensured using a rotating communication device mounted on a rotation axis of the moving unit. Multisegment assemblies of optical wedges are used instead of single wedges in the compensator for the tangential motion blur of the image. We demonstrate that this engineering design makes it possible to increase the signal integration time by several times compared with a basic design with single-segment wedges.

A motion blur compensation algorithm for 2D DIC measurements of deformable bodies

Digital image correlation (DIC) is a widely adopted optical technique to detect motions and deformations dealing with vision-based systems. Considering DIC dynamic applications, motion blur represents one of the most relevant sources of measurement uncertainty. This paper proposes an innovative approach to compensate the effect of motion blur on images coming from deformable bodies, where the amount of blur (w) varies within the single frame. A subset-based technique is developed to estimate and then remove motion blur for each image region. The algorithm is validated both on synthetic images and experimental data, demonstrating its effectiveness at improving the accuracy of the displacements measured by DIC.

Estimation and compensation of motion blur for the reduction of uncertainty in DIC measurements of flexible bodies

Digital image correlation (DIC) is a useful technique to measure displacement/strain fields both for static and dynamic problems in experimental mechanics. When monitoring moving objects with digital cameras, motion blur may occur if the shutter time reaches the time scale of the motion of the measurand. Consequently, motion blur is one of the most relevant problems in those dynamic DIC applications where shutter time cannot be set arbitrarily. This work deals with the problem of compensating motion blur effects on a generic DIC image. The problem of motion blur compensation to reduce DIC uncertainty is discussed in literature in the case of rigid target, where the amount of motion blur is the same for the whole image area. Deformable targets, instead, pose the additional problem when motion blur is variable within a single frame. In this paper a subset-based technique is proposed to estimate and compensate the motion blur for each image region. The approach is tested on synthetically deformed and blurred images of a notched beam specimen.

Off-axis virtual-image display and camera by holographic mirror and blur compensation.

We propose off-axis virtual-image display and camera systems, which integrate a vertically-standing holographic off-axis mirror, blur-compensation optical systems, and digital imaging devices. In the system, the holographic mirror is used for an off-axis reflector, which realizes an upright and thin screen for virtual-image formation. By combining it with a display unit, an off-axis virtual-image display is realized, where the virtual image can be seen behind the upright holographic mirror. Simultaneously, by combining it with a camera unit, an off-axis camera is implemented, which realizes frontal shooting of objects by a camera placed at an off-axis position. Since both the off-axis display and the camera can be implemented by a single holographic mirror, it can be applied to a two-way visual-telecommunication system with a thin screen, which implements eye contact and the observer--image distance. A problem with the proposed system is image blur, which is caused by the chromatic dispersion of the holographic mirror. To solve this, we designed optical blur-compensation systems using a diffractive optical element and a diffuser or a lens. Experimental results verify the concept of the proposed systems with clarifying the effect of designed blur-compensation methods.

Shape Sensing Method for Micro Fabricated Objects based on Real-time High-speed Motion Blur Compensation Method using Galvanometer Mirror

Shape Sensing Method for Micro Fabricated Objects based on Real-time High-speed Motion Blur Compensation Method using Galvanometer Mirror

Resolution enhancement in in-line holography by numerical compensation of vibrations

Mechanical vibrations of components of the optical system is one of the sources of blurring of interference pattern in coherent imaging systems. The problem is especially important in holography where the resolution of the reconstructed objects depends on the effective size of the hologram, which is on the extent of the interference pattern, and on the contrast of the interference fringes. We discuss the mathematical relation between the vibrations, the hologram contrast and the reconstructed object. We show how vibrations can be post-filtered out from the hologram or from the reconstructed object assuming a Gaussian distribution of the vibrations. We also provide a numerical example of compensation for directional motion blur. We demonstrate our approach for light optical and electron holograms, acquired with both, plane- as well as spherical-waves. As a result of such hologram deblurring, the resolution of the reconstructed objects is enhanced by almost a factor of 2. We believe that our approach opens up a new venue of post-experimental resolution enhancement in in-line holography by adapting the rich database/catalogue of motion deblurring algorithms developed for photography and image restoration applications.

Compensation for Blur Requires Increase in Field of View and Viewing Time.

Spatial resolution is an important factor for human pattern recognition. In particular, low resolution (blur) is a defining characteristic of low vision. Here, we examined spatial (field of view) and temporal (stimulus duration) requirements for blurry object recognition. The spatial resolution of an image such as letter or face, was manipulated with a low-pass filter. In experiment 1, studying spatial requirement, observers viewed a fixed-size object through a window of varying sizes, which was repositioned until object identification (moving window paradigm). Field of view requirement, quantified as the number of “views” (window repositions) for correct recognition, was obtained for three blur levels, including no blur. In experiment 2, studying temporal requirement, we determined threshold viewing time, the stimulus duration yielding criterion recognition accuracy, at six blur levels, including no blur. For letter and face recognition, we found blur significantly increased the number of views, suggesting a larger field of view is required to recognize blurry objects. We also found blur significantly increased threshold viewing time, suggesting longer temporal integration is necessary to recognize blurry objects. The temporal integration reflects the tradeoff between stimulus intensity and time. While humans excel at recognizing blurry objects, our findings suggest compensating for blur requires increased field of view and viewing time. The need for larger spatial and longer temporal integration for recognizing blurry objects may further challenge object recognition in low vision. Thus, interactions between blur and field of view should be considered for developing low vision rehabilitation or assistive aids.

Compensation for blur requires an increase in field of view

Compensation for blur requires an increase in field of view

Real-time high-speed motion blur compensation system based on back-and-forth motion control of galvanometer mirror.

We developed a novel real-time motion blur compensation system for the blur caused by high-speed one-dimensional motion between a camera and a target. The system consists of a galvanometer mirror and a high-speed color camera, without the need for any additional sensors. We controlled the galvanometer mirror with continuous back-and-forth oscillating motion synchronized to a high-speed camera. The angular speed of the mirror is given in real time within 10 ms based on the concept of background tracking and rapid raw Bayer block matching. Experiments demonstrated that our system captures motion-invariant images of objects moving at speeds up to 30 km/h.

Personal identification from degraded and incomplete high resolution palmprints

A high resolution palmprint recognition system using full or partial, eventually degraded, palmprints is presented. Previous work on palmprint matching addressed mostly commercial applications, using low resolution images. However, in forensic scenarios, high resolution palmprints, although incomplete and/or degraded, are often used. Degradations may result from surface irregularities or impurities, which are often modelled as Gaussian or salt and pepper noise, as well as smearing of the palmprint because of hand sliding, which in this work is modelled as motion blur. The proposed system matches palmprints, full or partial, undegraded or subjected to one of the above degradations, against palmprints registered in a database. The proposed system extends previous work of the authors by adaptively selecting between two palmprint matching approaches, achieving better recognition results than either of the two individual strategies. The first approach relies on a motion blur compensation technique, while the second is based on a combination of the Fourier–Mellin transform with a modified phase-only correlation matching strategy. The presented results show that for sufficiently large palmprint areas the blur compensation technique works better, while for small-sized partial palmprints with large motion blur degradation values the second approach based on correlation is preferred.

Myopia—Yesterday, Today, and Tomorrow

Ideas about the etiology of myopia used to be trapped in a debate of either/or. Early descriptions placed the blame on the environment, particularly long periods of close work.1 In the mid-20th century, Sorsby and Benjamin2 swung the pendulum strongly toward genetics when they stated flatly that “studies on twins have shown that all refractions of the eye—emmetropia, hypermetropia, and myopia—are determined genetically.” Nothing ruins a good dogma like data. The more recent view, informed by an explosion of information from research in the last 20 years, has moved toward a melding of these two extremes into a richly detailed combination of both nature and nurture. The idea that refractive error was hardwired into genes was shattered in 1977 when Wiesel and Raviola3 showed that eyelid suture produced myopia in monkeys and, then again, in 1978 when Wallman and coworkers4 showed that simple frosted occluders resulted in large amounts of myopia in chicks. The myopia research community truly lost one of its pioneers when Josh Wallman passed away in 2012. These deprivation experiments were followed by lens experiments by Schaeffel and coworkers5 whose powerful paradigm has shown that the eyes of young animals of an amazing variety of species respond to the sign and the magnitude of imposed defocus by modulating their rates of axial growth to compensate for the defocus and become effectively emmetropic. The fovea was originally assumed to be the seat of this local growth regulation because of its superior acuity, but later work showed that the retinal periphery also plays a role and can actually override the fovea during compensation for blur.6 This insight has stimulated interest in specialty refractive corrections such as multifocal contact lenses and orthokeratology, interventions that provide myopic defocus to the periphery in an attempt to rein in the growth of the myopic eye.

Erratum to: Prosthetic component segmentation with blur compensation: a fast method for 3D fluoroscopy

Erratum to: Prosthetic component segmentation with blur compensation: a fast method for 3D fluoroscopy

Prosthetic component segmentation with blur compensation: a fast method for 3D fluoroscopy

A new method for prosthetic component segmentation from fluoroscopic images is presented. The hybrid approach we propose combines diffusion filtering, region growing and level-set techniques without exploiting any a priori knowledge of the analyzed geometry. The method was evaluated on a synthetic dataset including 270 images of knee and hip prosthesis merged to real fluoroscopic data simulating different conditions of blurring and illumination gradient. The performance of the method was assessed by comparing estimated contours to references using different metrics. Results showed that the segmentation procedure is fast, accurate, independent on the operator as well as on the specific geometrical characteristics of the prosthetic component, and able to compensate for amount of blurring and illumination gradient. Importantly, the method allows a strong reduction of required user interaction time when compared to traditional segmentation techniques. Its effectiveness and robustness in different image conditions, together with simplicity and fast implementation, make this prosthetic component segmentation procedure promising and suitable for multiple clinical applications including assessment of in vivo joint kinematics in a variety of cases.

Generating realistic retinal image for whole visual system modeling

Recent progress and improvements on optical technology have enabled us to measure the characteristics of the eye in great detail. It has also been shown that, even when myopia and astigmatism are completely corrected with spectacle, the retinal image is still blurred due to uneven refraction power of the cornea and the crystalline lens [1][2], an effect known as “irregular astigmatism”. However, almost all researches in visual science assume sharp images as input. Currently, a large-scale whole visual system models is being developed on a supercomputer to elucidate its complex function [3]. Moreover, it has been demonstrated that the blurred retinal images are gradually compensated at both, the retina and the cortex [4]. Therefore, we have developed an eye optics model which can calculate the blurred retinal image based on known physiological evidences to provide a more realistic input to retinal and visual cortex model. The proposed model is described by the linear image filter based on Artal's model [5]. It can process multi-spectral image for reproducing the spectral characteristics of photoreceptors. OTF (Optical transfer function) calculated from the wavefront aberration with spectral transmittance of the lens is used for image processing filter. The wavefront aberration is defined by Zernike Coefficients values (measured by wavefront aberrometer such as OPD-Scan or Shack-Hartman Sensor systems) or SCA values (power of sphero-cylindrical lens with its axis for spectacle lens or contact lens prescriptions). To consider the effects of diffraction, OTF is designed to vary depending on pupil diameter and wavelength. To account for aging effects, the spectral transmittance of the lens [6] was also implemented. As inputs of the model, multispectral images taken by multispectral camera or converted from RGB image can be utilized. We simulated a closed-loop system connecting the proposed model to a simple eye movement model and pupil model. The eye movement model generates random saccades and the pupil model modifies pupil diameter according to the mean luminance of retinal image. As a result, our model could reproduce the effect of chromatic aberration, pupil diameter and age on the blurred retinal image. By applying our model to the large-scale visual system model, it is expected to be able to evaluate the phenomena, for example, age dependent perception on visual illusions, blur compensation in retina or visual cortex, and so on.

Neural compensation for long-term asymmetric optical blur to improve visual performance in keratoconic eyes.

Purpose. To investigate whether long-term visual experience with irregular optical blur compensates for the impact of higher-order aberration on visual performance in keratoconic (KC) eyes. Methods. The aberrations and high (100%)- and low (20%)-contrast tumbling E visual acuity (VA) were measured in four moderate KC eyes in which the subjects were wearing their own prescribed soft toric contact lenses over a 6-mm pupil. VA was measured in three emmetropic normal eyes for comparison with each of the four KC eyes. An adaptive optics system was used to correct the aberration of the normal eye and to induce the aberration of the KC eye simultaneously during vision testing. The magnitude of neural compensation was defined as improvement in VA in each KC eye compared with the normal eyes with KC aberrations. Results. Mean total and higher-order root mean square errors in the KC eyes with contact lenses were 2.72 +/- 0.83 mum and 1.36 +/- 0.29 mum, respectively, for a 6-mm pupil. Residual RMS wavefront error in induction of KC aberrations on normal eyes was approximately 0.1 mum in all cases. Each KC eye had statistically better high (P < 0.02)- and low (P < 0.03)-contrast VA than the three normal eyes. Mean compensation for high-contrast VA in logMAR was 0.12 +/- 0.09, corresponding to an improvement of 23.8%. A similar result was obtained for low-contrast VA. The magnitude of compensation increased with the severity of KC aberrations. Conclusions. In KC eyes, the neural visual system compensates for long-term visual experience with an asymmetrically blurred retinal image, resulting in improved visual performance.

Model-Based Blur Compensation and Shadow Removal in Multi-Projection Environment Based on Spatial Frequency Analysis of Projected Image

Model-Based Blur Compensation and Shadow Removal in Multi-Projection Environment Based on Spatial Frequency Analysis of Projected Image

Bi-ComForWaRD: BIVARIATE COMPLEX FOURIER-WAVELET REGULARIZED DECONVOLUTION FOR MEDICAL IMAGING

In this paper, we propose a new hybrid Bivariate Complex Fourier Wavelet Regularized Deconvolution (Bi-ComForWaRD) that is an extension to the ComForWaRD algorithm, for medical imaging. This new algorithm is a two-step process, a global blur compensation using generalized Wiener filter and followed by a denoising algorithm using local adaptive Bivariate shrinkage function. It is a low-complexity denoising algorithm using the joint statistics of the wavelet coefficients and considers the statistical dependencies between the coefficients. And also, the performance of this system will be demonstrated on both the orthogonal wavelet transform and the dual-tree complex wavelet transform (DT-CWT) and some comparisons with the best available wavelet-based image denoising results will be given in order to illustrate the effectiveness of the system.