Temporal feature datasets, today’s reality

By J. Michael Brown

Geographic information systems and location based services have the power to change the world through significant commercial and socioeconomic initiatives, so it is ironic that their global application and economic value are severely limited by the world's constant change. Decades of research and development have proven, purportedly beyond doubt, that true spatiotemporal map data — current, complete and accurate — is incompatible with low cost, and therefore, "impossible;" but new technology has emerged that demands we change our thinking and embrace change.

Born of frustration and a vision to change the world

Frustrated that expensive databases quickly lost value and usefulness in the absence of affordable updates, World of Change (WOC), Gainesville, FL, was founded in 2007 with a vision: to develop object-based change detection processes integrating crowdsourcing for complete, correct and ever-current geospatial data. WOC believed this would accelerate educational, governmental and commercial innovation that, in turn, would vigorously stimulate global economies by creating jobs, as well as improving standards of health and living.  WOC knew that fulfilling this lofty vision would require overcoming the real and perceived "impossible" incompatibility between automation (speed and low cost) and cognitive ability (completeness and correctness) for maintaining temporal updates. That challenge drove them to succeed.

Normalizing the impossible

WOC began by challenging the longstanding shortcomings of semi-automated processing – the point where direct opposing attributes of automation and human input progressively repel until objectives collapse at unacceptable levels of accuracy, timeliness and cost.  Envisioned were crossover processes that would essentially neutralize the repelling attributes and accentuate the desired results. WOC's ultimate goal was to pull change detection from the realm of the impossible into everyday utility.

Beginning with building outlines, early proof of concept success quickly led to rising success in real-world production contracts, ultimately becoming FinitEdge™ technologies.  FinitEdge™ combines unique image automation and dedicated crowd workforce classification processes to produce fast, high-accuracy feature extraction and change detection at an affordable cost, overcoming the barriers to worldwide map creation and near real-time updates.

The processes extract features from orthoimagery for new map creation, and analyze vector-to-ortho or ortho-to-ortho datasets for detecting new, demolished, and changed features to update an existing geodatabase.  

FinitEdge™ change detection and building polygon extraction using both low resolution and high compression imagery.

Change detection made easy

Once the complex challenges were identified and overcome, feature extraction and change detection processing became quite easy, especially for the consumer.  Input is as simple as orthophotography and a defined area of interest for feature extraction; change detection requires the same plus existing vector data. Users can change, alter or remove all geospatial references for location security and strip all attributes for privacy. The output is in ESRI shapefile format with the same reference frame as the input data.

Attaining accuracy and completeness

Accuracy and completeness is ensured through the analysis of each selected image segment by two to three independent crowd workforce analysts. Repetitive analysis seems counterintuitive to both speed and low cost, but the segmentation process drives the workforce through a highly efficient analysis that is as economical as it is comprehensive. Development and testing on hundreds of real-world projects has consistently yielded extraction and change detection of points, polylines and polygons on millions of buildings with clients reporting average false negative/positive accuracies exceeding 99.5%, and on more than forty thousand miles of road edges with no client rejections.

Meeting GEOINT security demands

FinitEdge™ was specifically designed to secure all sensitive and classified data for international use in commercial, government and military applications. It processes any size area from single images to entire countries and requires no geospatial reference frames.

To meet stringent levels of security, its finite edge detection capabilities for infrastructure — buildings, roads and impervious surfaces — use ungeoreferenced input and apply multiple levels of random encryption to highly segmented data, which is then indiscriminately distributed to a proprietary and totally disparate global crowd workforce.

Making it affordable

A multitude of applications exist utilizing FinitEdge™. In all cases, the ultimate goal is to drive prices as low as possible so that every nation is completely mapped and can afford to update as frequently as desired. This goal is achieved through:

  • efficient programming;
  • a highly scalable crowd workforce;
  • a forthcoming Software as a Service interface, allowing clients to manage their own budget priorities; and,
  • a privately owned corporation free from investor and stockholder pressures for profit.

The millions of buildings extracted and/or analyzed for change have, to date, cost far less than a postage stamp. This means that FinitEdge™ change detection could provide a national census agency with positional errors and omissions for every building structure point in their database for what it now costs to just mail letters announcing upcoming census surveys! A national census can now be executed with high accuracy and low cost, truly reflecting the technology of the times.

Cost/benefit sharing

This low cost capability to accurately map and maintain temporal data will hopefully encourage greater cooperation between commercial, governmental and educational entities — eventually making all feature layers a basic utility — where successful use and product competition is established by intelligent application of the base data rather than by its ownership. This game-changing attitude reduces duplication and overall cost, increases competition through application innovation, and profoundly benefits the health and wealth of communities worldwide.

For example, road/pavement edges have variable applications such as: visual representation on planimetric and topographic maps, pavement management, impervious surfaces management, centerline generation, LiDAR breaklines, traffic management, vehicle navigation, advanced driver assistance systems, and more. The socioeconomic benefits of road edge polygons is broad, but initial development and temporal maintenance of this data layer has previously been cost prohibitive, if not outright impossible, due to the undetected frequency of new construction, curb cuts, reconstruction of islands and turn lane storage, lane widening, roundabout additions, road realignment/straightening and so forth.

FinitEdge™change detection and feature extraction composite showing existing, new and abandoned edge of pavement.

Cost and imagery sharing between federal, state, local, commercial and private interests — to create a common feature layer providing for such applications — would leave forward-thinking commercial market leaders confident, accelerate innovation, inspire entrepreneurial opportunity, enhance local government asset management, stimulate local economies worldwide, improve health, alleviate poverty, reduce disaster risk and increase disaster response for relief, all internationally.

Observing socioeconomic impact

The road edge example demonstrates how cost sharing on one feature layer can result in a number of socioeconomic benefits. Similar benefit exists where data creation and maintenance is implemented through a single entity at the national or international level.

For example, there is a real need to have every building structure, mud hut or tent identified by Pintroids™, which are rooftop point features, whether for disaster preparedness, population census or military strategy. If a national government or the United Nations sponsored this point feature layer, then its creation and continuous maintenance would not only satisfy disaster, census and military needs at the national or international levels, but would also benefit applications ranging from regional and global commercial products to detecting new real property tax revenues.

FinitEdge™extraction of building structure Pintroids™ in urban, suburban, jungle and slum areas.

Celebrating job creation

Though some fear that semi-automation — automated segmentation with crowdsourcing analysis —will kill job opportunities in mapping and GIS, the opposite is true. Entrepreneurial opportunities will abound. Cost effective FinitEdge™ processes generate 2D vector data, creating endless new job opportunity for traditional geospatial mapping firms and agencies through quality assurance and control, data integration, stereo-enhancement, topology, intelligence, 3D conversion and more. This breakthrough releases an immeasurable amount of data creation and continual updating, requiring unending GIS/LBS mapping expertise to maintain the incalculable usage, value and benefit derived from these raw feature datasets.

Changing the playing field

Major players in map navigation and LBS could immediately use this authoritative output to validate and fill voids in existing data. Though each may initially choose to act alone to protect current content and market share, FinitEdge™ foreshadows the eventual sharing of common 2D feature data with a shift of resources to compete based on product intelligence, application and functionality more than on temporal map feature coverage and completeness.

Temporal feature datasets are today's reality – it's time to embrace change!

If you liked this article subscribe to our newsletter...stay informed on the latest geospatial technology

© 2016 Directions Media. All Rights Reserved.