When designing computer vision systems for the blind and visually impaired it is important to determine orientation of user relative scene. We observe that most indoor outdoor (city) scenes are designed on a Manhattan three-dimensional grid. This grid structure puts strong constraints intensity gradients in image. demonstrate an algorithm detecting such based Bayesian inference using statistics which we have learnt this domain. Our requires single input image does not involve pre-processing...

We formulate edge detection as statistical inference. This is data driven, unlike standard methods for which are model based. For any set of filters (implementing local cues), we use presegmented images to learn the probability distributions filter responses conditioned on whether they evaluated or off an edge. Edge formulated a discrimination task specified by likelihood ratio test responses. approach emphasizes necessity modeling image background (the off-edges). represent conditional...

How do we decide where to look next? During natural, active vision, move our eyes gather task-relevant information from the visual scene. Information theory provides an elegant framework for investigating how stimulus combines with prior knowledge and task goals plan eye movement. We measured movements as observers performed a shape-learning -matching task, which was tightly controlled. Using computational models, probe underlying strategies used by when planning their next One strategy is...

Computer vision holds the key for blind or visually impaired to explore visual world.

This letter argues that many visual scenes are based on a “Manhattan” three-dimensional grid imposes regularities the image statistics. We construct Bayesian model implements this assumption and estimates viewer orientation relative to Manhattan grid. For images, these good approximations (as estimated manually by authors). These also make it easy detect outlier structures unaligned To determine applicability of world model, we implement null hypothesis assumes statistics independent any...

In this paper we propose a computer vision-based technique that mines existing spatial image databases for discovery of zebra crosswalks in urban settings. Knowing the location is critical blind person planning trip includes street crossing. By augmenting (such as Google Maps or OpenStreetMap) with information, traveler may make more informed routing decisions, resulting greater safety during independent travel. Our algorithm first searches satellite images; all candidates thus found are...

We treat the problem of edge detection as one statistical inference. Local cues, implemented by filters, provide information about likely positions edges which can be used input to higher-level models. Different cues evaluated effectiveness their corresponding filters on a dataset 100 presegmented images. use theoretic measures determine variety different detectors working at multiple scales black and white color Our results give quantitative for advantages multi-level processing,...

Urban intersections are the most dangerous parts of a blind or visually impaired person's travel. To address this problem, paper describes novel "Crosswatch" system, which uses computer vision to provide information about location and orientation crosswalks pedestrian holding camera cell phone. A prototype system runs on an off-the-shelf Nokia N95 phone in real time, automatically takes few images per second, analyzes each image fraction second sounds audio tone when it detects crosswalk....

We describe a wayfinding system for blind and visually impaired persons that uses camera phone to determine the user's location with respect color markers, posted at locations of interest (such as offices), which are automatically detected by phone. The marker signs specially designed be in real time cluttered environments using computer vision software running on phone; novel segmentation algorithm quickly locates borders each image, allows calculate how far is from present model scanning...

For blind travelers, finding crosswalks and remaining within their borders while traversing them is a crucial part of any trip involving street crossings. While standard Orientation & Mobility (O&M) techniques allow travelers to safely negotiate crossings, additional information about other important features at intersections would be helpful in many situations, resulting greater safety and/or comfort during independent travel. instance, planning pedestrian may wish informed the...

Smartphone apps can use object recognition software to provide information blind or low vision users about objects in the visual environment. A crucial challenge for these is aiming camera properly take a well-framed picture of desired target object. We investigate effects two fundamental constraints -- frame rate and field view on person's ability an smartphone app. The app was used by 18 participants find targets beyond arm's reach approach them within 30 cm. While we expected that faster...

This paper describes recent progress on the "Crosswatch" project, a smartphone-based system developed for providing guidance to blind and visually impaired travelers at traffic intersections. Building past work Crosswatch functionality help user achieve proper alignment with crosswalk read status of walk lights know when it is time cross, we outline directions now taking realize its potential becoming practical system: namely, augmenting computer vision other information sources, including...

Although AR technology has been largely dominated by visual media, a number of tools using both and auditory feedback have developed specifically to assist people with low vision or blindness - an application domain that we term Augmented Reality for Visual Impairment (AR4VI). We describe two AR4VI at Smith-Kettlewell, as well pre-existing examples. emphasize is powerful tool the potential remove significantly reduce range accessibility barriers. Rather than being restricted use impairments,...

Wayfinding is a major challenge for visually impaired travelers, who generally lack access to visual cues such as landmarks and informational signs that many travelers rely on navigation. Indoor wayfinding particularly challenging since the most commonly used source of location information wayfinding, GPS, inaccurate indoors. We describe computer vision approach indoor localization runs real-time app conventional smart-phone, which intended support full-featured in future will include...

Modern household appliances, such as microwave ovens and DVD players, increasingly require users to read an LED or LCD display operate them, posing a severe obstacle for persons with blindness visual impairment. While OCR-enabled devices are emerging address the related problem of reading text in printed documents, they not designed tackle challenge finding characters appliance displays. Any system these must first locating among substantial amounts background clutter; moreover, poor...