Robert Colwell Musings Part 2

I started this last year when I was working on a retrospective of remote sensing of forests in California for the centennial of Berkeley Forestry. In the article I tried to highlight some of the pioneering work done by remote sensors that focused on Californian forests from the 1960s through the use of lidar today. As is often the case, the paper changed into something a bit more focused on lidar technology, and I had to cut most of the Colwell stuff. So, I reprise some information about him here, add my perspective on his work as antecedants to modern OBIA approaches, and include his rad interior design ideas. 

Early musings on forest objects by Colwell in 1964.

Dr. Colwell was an internationally renowned remote sensing scientist; he was former associate director of the Space Sciences Laboratory at the UC Berkeley, and he was the instructor of remote sensing in our own Mulford Hall from 1947 until his retirement in 1983. He was NASA co-investigator for Apollo IX, and his research in the 1960s on reflectance and multispectral reconnaissance were the primary basis for selecting the type of sensors and the spectral bands implemented in Landsat. Neat guy, and we all benefit from his intellectual legacy.

Finding meaningful blobs - a geographer’s quest

In our new Taylor and Francis book chapter, Thomas Blaschke, Helena Merschdorf, and I discuss Object-Based Image Analysis: Evolution, History, State of the Art, and Future Vision (Book website).  I did some work looking into Colwell's work, and found lots of discussion of nascent work describing object based approaches to image analysis. He struggled with the inability of algorithms to pull from digital imagery meaningful "blobs". See the examples here. 

One of Colwell's early attempts at object based work from 1973. Printout on a dot matrix printer?

His assessment of the potential for automation of an object recognition process depended on the capacities of a digital scanner and the ability of an algorithm to assess the differences, in photographic tone, between a "blob" and its surroundings (Colwell 1964, 1965). Colwell was an important advisor on the Landsat 1 mission, and his ideas on extraction of meaningful features transferred to his ambitions for the satellite missions (Colwell 1973). 

Maps as hipster decorations

I read some of his work as he transitioned from aerial photography to digital imaging, and I came across this picture. Mulford is just off the scene in the upper left corner, and Hearst Gym pool is visible in lower part. 

Oblique aerial view of Berkeley Campus of University of California taken with Camouflage Detection film. Robert N. Colwell

In his caption he says:

"Oblique aerial view of Berkeley Campus of University of California taken with Camouflage Detection film."(That is what they used to call color infrared.) "Such photography is superior to any other for certain photo interpretation purposes as indicated by some of the preceding examples. Note in this photo how color values for each species of tree tend to remain uniform from foreground to background because of the superior haze penetration offered by this film. The relatively long wavelengths to which this infrared-sensitive film reacts are scattered but very little by atmospheric haze particles, thus accounting for the uniform color values and for excellent image sharpness." I dig this part: "The original color transparencies have the same color values as seen here and consequently make very attractive panels for lamp shades, although certain of their colors fade upon prolonged exposure to light."

The trend for using map products as kitchy home decorations PRE-DATES 1970! Take that hipsters!

Articles mentioned:

Colwell, R.N. 1964. Aerial photography - A valuable sensor for the scientist. American Scientist, Vol. 52, No. 1 (MARCH 1964), pp. 16-49

Colwell, R.N., 1973. Remote Sensing as an Aid to the Management of Earth Resources. American Scientist. 61(2): 175-183.

Some more about him here:

Launch of Sonoma County Veg Mapping Program

The Sonoma County Agricultural Preservation and Open Space District has begun a 3-5 year program to map Sonoma County’s diverse plant communities.

An accurate, up-to-date map of vegetation and habitat type is key to ensuring good planning and management for watershed protection, flood control, fire and fuels management, and wildlife habitat conservation. A vegetation map is also critical to assessing climate benefits provided by the landscape, such as the amount of carbon being absorbed from the atmosphere or the degree to which the landscape is buffering extreme weather events.

These folks are using 3-6-inch CIR imagery and obia to map vegetation across Sonoma County. GIF is serving up the imagery! Check it out!

Definiens Earth Science bought by Trimble

The new company will be called Trimble Geospatial Munich, and maintain the same staff, and the Centers of Excellence Program.  From the Open Letter:

"Definiens’ earth sciences business, including the eCognition suite of products, was acquired on June 10, 2010 by Trimble. The eCognition team has now transferred to Trimble, and the business of delivering the most advanced geospatial analysis software will continue, uninterrupted.

This change represents a significant step forward for eCognition. Access to Trimble’s advanced technologies, expertise and global operations provides us with an exceptional opportunity to take eCognition to a new level. In time, the benefits of the synergies between Trimble and eCognition will become evident through the exciting new product and service innovations we deliver together.

It is also worth noting that the collaboration with Definiens does not end with this transaction. In fact, Definiens and Trimble have signed a co-development agreement to ensure that the core technology driving eCognition continues to evolve and improve, and that there is no pause in the development or release schedule for eCognition."

Using LiDAR las files in next eCognition version

via Andreas Lang at the Definiens Community

Can we load and process LiDAR las files in Definiens eCognition (Developer or Server) directly?

The new Definiens software will have two ways for handling las files via converting them into rasters directly in the Software:

  • a raster driver for loading and visualizing these kind of images (with an appropriate dialog for setting the resolution for converting the point cloud to a 2D raster) using the driver the user can see the intensity data and select an appropriate subset;
  • an algorithm for converting the existing loaded image layer (las file) into a feasible layer with appropriate data of intensity, elevation, class or number of returns for further processing with much more functionality for filtering:
    • By Return (All/First/All)
    • By Classes
    The user can also select the kind of calculation for a raster cell value (Average, Minimum, Maximum, Median, Most frequently. value).


Definiens community site launched

At our Definiens workshop today, Juan mentioned that the Definiens community website has been launched in the last month. Definiens is one of the few comprehensive segmentation and classification software solutions targeted for high spatial resolution imagery processing.  I've used the software for mapping tidal wetland sites, and for mapping dead crowns in oak forests.  The community site includes lots of information on shared codes, videos, demos, and discussion.  Since the software is so comprehensive, and has so many possible options, it can be difficult to become an expert. I personally like to know the "choice domain" within which I am working, so learning Definiens places me somewhere on the edge of my comfort zone. But with this, especially the rule set sharing, I think I can get stuck in. Thanks Juan!

The GIF is one of 7 Centers of Excellence for Definiens; other sites include: