Eli Peli

Contrast in complex images.

The physical contrast of simple images such as sinusoidal gratings or a single patch of light on a uniform background is well defined and agrees with the perceived contrast, but this is not so for complex images. Most definitions assign a single contrast value to the whole image, but perceived contrast may vary greatly across the image. Human contrast sensitivity is a function of spatial frequency; therefore the spatial frequency content of an image should be considered in the definition of contrast. In this paper a definition of local band-limited contrast in images is proposed that assigns a contrast value to every point in the image as a function of the spatial frequency band. For each frequency band, the contrast is defined as the ratio of the bandpass-filtered image at the frequency to the low-pass image filtered to an octave below the same frequency (local luminance mean). This definition raises important implications regarding the perception of contrast in complex images and is helpful in understanding the effects of image-processing algorithms on the perceived contrast. A pyramidal image-contrast structure based on this definition is useful in simulating nonlinear, threshold characteristics of spatial vision in both normal observers and the visually impaired.

Image enhancement using a contrast measure in the compressed domain

An image enhancement algorithm for images compressed using the JPEG standard is presented. The algorithm is based on a contrast measure defined within the discrete cosine transform (DCT) domain. The advantages of the psychophysically motivated algorithm are 1) the algorithm does not affect the compressibility of the original image because it enhances the images in the decompression stage and 2) the approach is characterized by low computational complexity. The proposed algorithm is applicable to any DCT-based image compression standard, such as JPEG, MPEG 2, and H. 261.

Reorganization of Visual Processing in Macular Degeneration

Macular degeneration (MD), the leading cause of visual impairment in the developed world, damages the central retina, often obliterating foveal vision and severely disrupting everyday tasks such as reading, driving, and face recognition. In such cases, the macular damage eliminates the normal retinal input to a large region of visual cortex, comprising tens of square centimeters of surface area in each hemisphere, which is normally responsive only to foveal stimuli. Using functional magnetic resonance imaging, we asked whether this deprived cortex simply becomes inactive in subjects with MD, or whether it takes on new functional properties. In two adult MD subjects with extensive bilateral central retinal lesions, we found that parts of visual cortex (including primary visual cortex) that normally respond only to central visual stimuli are strongly activated by peripheral stimuli. Such activation was not observed (1) with visual stimuli presented to the position of the former fovea and (2) in control subjects with visual stimuli presented to corresponding parts of peripheral retina. These results demonstrate large-scale reorganization of visual processing in MD and will likely prove important in any effort to develop new strategies for rehabilitation of MD subjects.

A method for objective edge detection evaluation and detector parameter selection

Subjective evaluation by human observers is usually used to analyze and select an edge detector parametric setup when real-world images are considered. We propose a statistical objective performance analysis and detector parameter selection, using detection results produced by different detector parameters. Using the correspondence between the different detection results, an estimated best edge map, utilized as an estimated ground truth (EGT), is obtained. This is done using both a receiver operating characteristics (ROC) analysis and a Chi-square test, and considers the trade off between information and noisiness in the detection results. The best edge detector parameter set (PS) is then selected by the same statistical approach, using the EGT. Results are demonstrated for several edge detection techniques, and compared to published subjective evaluation results. The method developed here suggests a general tool to assist in practical implementations of parametric edge detectors where an automatic process is required.

Field expansion for homonymous hemianopia by optically induced peripheral exotropia

Purpose To describe a novel method for prism correction of hemianopia that provides field-of-view expansion in a convenient and functional format and to evaluate initial clinical application. Method To expand the upper quadrant of the field, a high power prism segment (30–40&Dgr;) is placed base-out across the upper part of the spectacle lens, on the side of the loss, at about the level of the limbus. A similar prism segment at the lower part of the lens is used to treat the lower field. The peripheral location of the prisms causes peripheral exotropia. As a result a scene segment as high as the vertical span of the prism is shifted laterally by 15 to 20° relative to the view of the other eye. At the edge of the hemianopic field loss, objects that would fall in the scotoma of one eye are seen through the prism in the other eye, providing a simultaneous awareness of details within the otherwise absent field-of-view. An approach for fitting the system to patients with abnormal binocular vision (strabismus and amblyopia, with or without diplopia) is discussed as well. The effect of the prisms was evaluated in a noncomparative case series (12 patients). Results The field expansion is provided at any position of lateral gaze, including gaze away from the side of the scotoma. The effect of this technique on field expansion was demonstrated using standard binocular perimetry. Most patients reported substantial improvement in function and in obstacle avoidance. Conclusion A novel method for the optical treatment of hemianopia was developed and tested. It was found to be effective in expanding the field and helping patients’ mobility.

The visual effects of head-mounted display (HMD) are not distinguishable from those of desk-top computer display

Concerns about potentially harmful effects on the visual system due to the use of head mounted displays (HMDs) in general, and stereoscopic systems in particular, have been raised in the literature. Most of the concerns were based on studies measuring visual function changes following short-term use of HMDs. This study measured functional changes in binocular vision, accommodation, and resolution following 30 min use of HMD in both stereoscopic- and non-stereoscopic modes, and compared them to changes following the same task performed on a desk-top CRT display. No functional differences were found between HMD and CRT and most measured changes were too small to be considered clinically meaningful. An evaluation of subjective comfort found a statistically significant difference in the impression of comfort between the CRT and the HMD in stereoscopic mode, with the latter being less comfortable. It can be concluded that the functional changes reported following short term use of HMDs are not specific to stereoscopic presentation and do not differ from those caused by desk-top CRT display.

Feature-Based Registration of Retinal Images

Registration of retinal images taken at different times frequently is required to measure changes caused by disease or to document retinal location of visual stimuli. Cross-correlation has been used previously for such registration, but it is computationally intensive. We have modified a faster algorithm, sequential similarity detection (SSD), to use only the portion of the template that contains retinal vessels. When compared to standard SSD and cross-correlation, this modification improves the reliability of detection for a variety of retinal imaging modalities. The improved reliability enables implementation of a two-stage registration strategy that further decreases the amount of computation and increases the speed of registration.

Image enhancement for the visually impaired: the effects of enhancement on face recognition

Image enhancement has been shown to improve face recognition by visually impaired observers. We conducted three experiments in an effort to refine our understanding of the parameters leading to this effect. In experiment 1 we found that the band of spatial frequencies between 4 and 8 cycles/face is critical for face recognition. In experiment 2 we found that enhancement of these frequencies and the resulting image distortion actually reduced recognition performance for normal observers. Since the degradation of performance by low vision is larger than the effect of distortion, the enhancement that reduces performance for normal observers may still be beneficial for the visually impaired observer. Experiment 3 found that patients tend to prefer images enhanced at frequencies higher than the critical frequencies found in experiment 1. Such individually selected enhancement did not improve recognition in comparison with uniformly applied enhancement. The lack of an enhancement effect may be due to the small variability in enhancement frequencies selected by our subject population.

Image Enhancement For The Visually Impaired

Application of image processing for the visually impaired is discussed. Image degradation in the low vision patients visual system can be specified as a transfer function obtained by measurements of contrast sensitivity. The effectiveness of adaptive image enhancement for printed pictures is demonstrated using an optically simulated cataractous lens.

On-road driving with moderate visual field loss

Purpose: We examined the relationship between visual field extent and driving performance in an open, on-road environment using a detailed scoring method that assessed the quality of specific skills for a range of maneuvers. The purpose was to determine which maneuvers and skills should be included in future, larger scale investigations of the effect of peripheral field loss on driving performance. Methods: Twenty-eight current drivers (67 ± 14 years) with restricted peripheral visual fields participated. Binocular visual field extent was quantified using Goldmann perimetry (V4e target). The useful field of view (UFOV®) and Pelli-Robson letter contrast sensitivity tests were administered. Driving performance was assessed along a 14-mile route on roads in the city of Birmingham, Alabama. The course included a representative variety of general driving maneuvers, as well as maneuvers expected to be difficult for people with restricted fields. Results: Drivers with more restricted horizontal and vertical binocular field extents showed significantly (p ≤ 0.05) poorer skills in speed matching when changing lanes, in maintaining lane position and keeping to the path of the curve when driving around curves, and received significantly (p ≤ 0.05) poorer ratings for anticipatory skills. Deficits in UFOV performance and poorer contrast sensitivity scores were significantly (p ≤ 0.05) correlated with overall driving performance as well as specific maneuver/skill combinations. Conclusions: In a small sample of drivers, mild to moderate peripheral visual field restrictions were adversely associated with specific driving skills involved in maneuvers for which a wide field of vision is likely to be important (however most were regarded as safe drivers). Further studies using similar assessment methods with drivers with more restricted fields are necessary to determine the minimum field extent for safe driving.