IMAGING STANDARDS

Solicitation for comments on

Solicitation for comments on SERNEC live plant image standards

Logistics:

1. This is a solicitation for comments regarding proposed standards for photographing live plant images. These standards are outlined at:
http://www.cas.vanderbilt.edu/bioimages/image-standards-frame.htm
2. These standards are being considered for adoption for use in a live plant image collection of SERNEC (the Southeast Regional Network of Expertise and Collections). They are intended to guide the collection of new images to be included in the SERNEC collection. Extant images that don't meet the standards may still be included in a "legacy" SERNEC collection.
3. The standards for herbaceous angiosperms, woody angiosperms, and gymnosperms are fairly "mature" and probably close to ready for adoption, in contrast to the cacti and fern standards which probably need to be looked at by specialists in those groups before they are ready to be adopted. So at this point, let's limit the discussion to the first three categories.
4. Please make initial comments by Nov. 1, 2008. The comments will be compiled for further discussion by those who wish to do so. Please email comments to steve.baskauf@vanderbilt.edu and indicate the question to which you are responding.

Considerations:

1. The standards are intended not only to produce standardized orientations of plant parts and to ensure that detailed images are collected of the features that are important for identification, but to facilitate rapid image collection under field conditions and to minimize the need for image manipulation during later processing.

2. Our primary purpose is to agree on standards for images that will be used to facilitate species identification and recognition and to be used as an educational resource. If there are competing considerations between these uses and others such as artistic or commercial uses, the educational considerations trump the others.

3. We are talking about the images that will be archived to serve as raw materials for other uses. We want to preserve the capability for these images to be used for as many other uses as possible. So even though modifications of the images such as creating composites of several features, writing the species name on the image itself, adding borders, etc. may be desirable for some uses, they would preclude other uses and are therefore not appropriate. Such modifications can be made on secondary products (e.g. publications and websites) that use the archive as a source.

Questions:

1. Background of images. In the images standards paper [Baskauf, S.J. and B.K. Kirchoff (2008) Digital plant images as specimens: toward standards for photographing living plants. Vulpina 7:16-30.], for close-up images we suggested using the combination of narrow aperture (i.e. high f stop), high shutter speed, and flash to allow for wide depth of field (all details in focus), reduced blurring due to motion of a hand-held camera, photography under low ambient light, and to produce a dark background that reduces distraction. See http://www.cas.vanderbilt.edu/bioimages/considerations.htm#flash for examples.

A. Are there reasons (practical or philosophical) NOT to do this?

B. Should the background be white instead of black, and how can that be achieved?
See http://www.cas.vanderbilt.edu/bioimages/considerations.htm#color for an example.

C. Are there enough advantages of natural lighting to override the advantages of flash? What are those advantages and what are the costs in terms of time and equipment to take good natural-light images?

2. Non-self-supporting objects. If it is not possible to photograph an object that will support itself (e.g. the only fruit available to be photographed is on the ground), how should the object be photographed? Holding it in the hand is easy, but has a distracting hand in the picture. Putting it on a black background reduces distraction, but requires carrying the background with you and has potential glare problems. Photographing the object on the ground looks more natural, but may produce a distracting background. See http://www.cas.vanderbilt.edu/bioimages/considerations.htm#fruit for examples.

3. Twigs and buds. The provisional standards specify that twigs be oriented horizontally (because that is generally the way they appear on the tree) and that closeup images of buds/leaf scars on winter twigs be oriented vertically (because that's usually the way they appear in line drawings in twig keys). See http://www.cas.vanderbilt.edu/bioimages/image-standards-woodangio.htm#twig

A. Does it make sense to have these orientations or would it be better to have all orientations be vertical with the apex to the top?

B. If the overall twig orientation remains horizontal, should it be specified that the apex always be in a particular direction (e.g. toward the right) or does this not matter?

C. Because winter buds are an important identifying characteristic and can be observed with a hand lens, they could be an important part of an image collection. However, because they will probably have to be photographed under a microscope, they fall beyond the general category of features that can be rapidly photographed in the field with a hand-held camera. Are they an important enough feature that they should be kept in the standards or are they a specialized feature that falls beyond the scope of this project?

D. Are there too many gymnosperm twig/needle categories? Images which show one feature often show the other as well (e.g. needle bases and twig showing attachment of needles - can some categories be combined? See http://www.cas.vanderbilt.edu/bioimages/image-standards-gymno.htm#twig

4. Seeds. Seeds are often small enough that they cannot be photographed well without using a microscope. They also may need to be extracted from the fruit. If small and outside of the fruit, they become non-self-supporting objects. See:
http://www.cas.vanderbilt.edu/bioimages/image-standards-woodangio.htm#open
http://www.cas.vanderbilt.edu/bioimages/image-standards-woodangio.htm#seed
http://www.cas.vanderbilt.edu/bioimages/image-standards-herb.htm#seeds

A. Are seeds a taxonomically important enough feature to be included in the standards as an object themselves to be photographed outside of the context of the fruit?

B. If seed photography is included in the standards, how should they be photographed?

C. If seed images generally need to be made microscopically, does this put them outside the scope of this project (i.e. images that can be taken in the field rapidly with a hand-held camera)?

5. Leaf orientation on herbs. The provisional standards have upper stem leaves oriented with the apex upward and basal/lower stem leaves oriented with apex downward. This was chosen because that seemed to most closely mimic their typical orientation on the plant. See
http://www.cas.vanderbilt.edu/bioimages/image-standards-herb.htm#leaf

A. Would it be better to have all of the herb leaf images oriented in the same direction (e.g. apex down as is the case with woody leaves) or should the standard remain as it is?

6. Habit/habitat shots. There is no specific standard indicating that the plant should be photographed in the context of its environment (e.g. a distant shot for small plants). See
http://www.cas.vanderbilt.edu/bioimages/image-standards-herb.htm#habitat

A. Is this an important-enough view that it should be included as a separate standard, or should it just be an optional addition to the whole plant image category?

7. Superfluous, missing, or inadequate standards.

A. Are there standards that you feel are unnecessary?

B. Are there standards that are not on the list that you think should be added?

C. What standards are in need of change?

Steve Baskauf - Vanderbilt Dept. of Biological Sciences
steve.baskauf@vanderbilt.edu

Summary of second round of

Summary of second round of discussion (December 5, 2008)
Compiled by Steve Baskauf

Live Plant Imaging standards: What do we photograph, how do we photograph it, and what kind of information do we want to include with the images?

I. Considerations:
A. Purpose. The standards are intended not only to produce standardized orientations of plant parts and to ensure that detailed images are collected of the features that are important for identification, but to facilitate rapid image collection under field conditions and to minimize the need for image manipulation during later processing.
B. Standardization of views. The development of standardized categories of features to be photographed ("views") serves several purposes:
1. It allows a user to search for similar kinds of images of features from different taxa. The views should be defined in the broadest way that allows functionally equivalent features to come up in a search (e.g. a search for "leaves" should return broadleaf leaves, conifer needles, and fern fronds because they are all functionally leaves vs. defining different views for each).
2. Minimizing the number of views of a particular feature (leaf, flower, etc.) to a few standardized views decreases the likelihood that images that are really of the same type are not found because their views are described in slightly different ways.
3. The creation of standardized views provides guidance for photographers so that they know what features are important to be photographed to systematically record the morphology of the plant.
C. Modifications to images. We are talking about the images that will be archived to serve as raw materials for other uses. We want to preserve the capability for these images to be used for as many other uses as possible. So even though modifications of the images such as creating composites of several features, writing the species name on the image itself, adding borders, etc. may be desirable for some uses, they would preclude other uses and are therefore not appropriate. Such modifications can be made on secondary products (e.g. publications and websites) that use the archive as a source. Likewise, metadata can be recorded in the EXIF data for the image or included in the SERNEC database and therefore do not need to be added to the visual images themselves.
D. Priorities. Our primary purpose is to provide standards for images that will be used to facilitate species identification and recognition and to be used as an educational resource. If there are competing considerations between these uses and others such as artistic or commercial uses, the educational considerations trump the others.

II. Standardized views
A. A set of standardized views should be defined for each major category of plants that have fundamentally different features. At this point (Dec. 2008) the views for woody angiosperms, generic herbaceous angiosperms, and conifers are "mature" enough to be considered as proposed SERENEC standardized views 1.0 . Views for cacti and ferns probably need more work. Views for grasses, sedges, composites, and mosses are needed but have not yet been developed.
B. Any views that are common to several feature categories should be defined as a single view. For example, there should be a single view for bark of a large tree, even though it will be a part of the sets of views for both woody angiosperms and conifers. Thus a non-redundant list of all accepted views will be smaller than the sum of views for all of the major categories.
C. List of views.
1. A list of proposed SERNEC Standardized Views 1.0 is posted at http://www.sernec.org/?q=node/213
2. Collections illustrating the characters have been released on the Morphbank website:
woody angiosperms: http://www.morphbank.net/myCollection/index.php?collectionId=464110
herbaceous angiosperms: http://www.morphbank.net/myCollection/index.php?collectionId=464267
gymnosperms: http://www.morphbank.net/myCollection/index.php?collectionId=464328
3. More images illustrating the standardized views are at:
http://www.cas.vanderbilt.edu/bioimages/image-standards-frame.htm
4. The standardized views were based on Baskauf, S.J. and B.K. Kirchoff (2008) Digital plant images as specimens: toward standards for photographing living plants. Vulpina 7:16-30.
http://www.cals.ncsu.edu/plantbiology/ncsc/vulpia/pdf/Baskauf_&_Kirchoff_Digital_Plant_Images.pdf

D. Based on comments made in the discussion, several changes were made from the standardized views suggested in Baskauf and Kirchoff (2008)
1. The view of the fascicle base (III.D.2) was very similar to the view of the twig showing attachment of needles (III.C.2) and was dropped from the list.
2. Seeds are now considered as a separate major category from fruit/cone. This will allow for searches of seed images across the major categories rather than differentiating between seeds from fruits and seeds from cones. Additional specifications or categories for seeds may need to be added after we gain more experience photographing them or receive input from others who already have that experience.
3. Because many monoecious or dioecious plant taxa have male and female inflorescences that differ greatly in appearance, separate subcategories were created for the whole inflorescence subcategory: whole inflorescence (unspecified), whole inflorescence (female), and whole inflorescence (male). Such a split was not implemented for other subcategories of flower views, but that could be considered in the future if enough images of male and female flowers and parts were included in the collection to warrant the split.
4. There was no clear consensus as to whether a view of the plant in its habitat should be added as a view separate from a general view of the whole plant, so it was not added as a subcategory. If necessary, it could added in the future.

III. Photographic techniques
A. Image background.
1. There was general agreement that distracting backgrounds (e.g. leaves on the forest floor) were not good for close-ups and that out-of-focus or single-color backgrounds were preferable.
2. In the discussion of image background, there were no compelling philosophical reasons raised for preferring black background over white, although a black background may be easier on the eyes while viewing and more esthetically pleasing. Because it requires no additional work to produce a black background when taking close-ups with a flash, black backgrounds are recommended as the default.
3. When photographing non-self-supporting objects (e.g. seeds) it is preferable to place them on a black background. High-quality black velvet was recommended. Photographing the seeds in situ in the fruit was also recommended. The most important consideration was avoiding a distracting background.

B. Natural light vs. flash.
1. It was noted that natural light produced more aesthetically pleasing images, but that may not outweigh the disadvantages of hauling around a tripod in the field.
2. Use of a flash, high shutter speed, and narrow aperture for close-ups provides a long depth of field, little blurring from motion, and a black background.
3. Due to lack of a clear consensus on this issue, it is recommended that photographers use the technique that seems the most appropriate for the situation.

IV. Metadata collected with the images
A. Date and time.
1. Information recorded automatically by camera.
B. GPS data
1. GPS data should be collected to document the location of all images.
2. This can be done manually using a regular GPS receiver, but specialized receivers and software can now embed the GPS data directly into the EXIF information included with the image.
3. The Darwin Core InformationWithheld element should be used to flag specimens of special concern that should have their GPS resolution reduced to prevent poaching or vandalism.
C. Scale information
1. If the number of pixels per unit length on the object is determined and archived with the image, software can be used to create scale bars or to allow distance measurements on the image.
2. Using only the distance from the object to the image plane, it should be possible with appropriate calculations to determine the number of pixels per unit length. Photographers should attempt to record this distance (either by reading it off of their lens or estimating it directly). We will work to come up with an accurate method for the calculation.
3. For the sake of simplicity, the standard unit length should be mm and the distance should be reported in mm.
D. Taxonomy, location (text), photographer, etc.
1. Recorded by photographer using any system desired.
E. Color scale. No opinions were expressed on this topic. Therefore the photographers should simply do their best to adjust the color balance of their images to match the color of the subject.

V. Image and metadata requirements
A. Meeting the standards. Taking images of all of the suggested views and recording all of the metadata listed is a tall order and probably few if any photographers would actually be able to do so. In addition, some photographers may not want to follow the recommendations listed here. What is the minimum level of conforming to the standards that is required for participation in the SERNEC live plant image collection? We are envisioning a two tier system: a collection of images that meet the full standards and a broader legacy collection that don't meet the full standards.

B. Minimum requirements for images included in the legacy collection:
1. The specimen is accurately identified to species.
2. The image can be categorized into one of the standard views and approximately meets the specifications of that view.
3. The copyright holder (i.e. photographer or institution) is willing to allow use of the image under a license no more restrictive than Creative Commons Attribution-Noncommercial-Share Alike 3.0 (BY-NC-SA). (See notes from previous discussion for more on this.)

C. Preferences for images included in the legacy collection.
1. Images from the same specimen are associated with each other.
2. Text location data (DwC Continent, Country, StateProvince, County, and Locality elements) provided.
3. High resolution image of at least 3 megapixels provided.

D. Minimum requirements for images included in the regular collection.
1. Images of enough views are taken of the specimen to allow for identification and annotation from the images. Images of the same specimen are associated with each other. The more views taken the better, but we recognize the time limitations of the photographer. The Baskauf and Kirchoff (2008) paper suggests which views are the most important to photograph. The specimen is accurately identified to species.
2. The image can be categorized into one of the standard views and meets the specifications of that view.
3. Text and GPS location data provided for the specimen. DwC InformationWithheld element should be used to flag sensitive species for which the GPS data reduced in resolution.
4. Date and time information are available in the EXIF data associated with the image.
5. The copyright holder (i.e. photographer or institution) is willing to allow use of the image under a license no more restrictive than Creative Commons Attribution-Noncommercial-Share Alike 3.0 (BY-NC-SA). See previous discussion for more on this.
6. High resolution image of at least 3 megapixels provided.
7. No distracting objects in image and no text added to image. No collages.

E. Preferences for images included in the regular collection.
1. High resolution image of at least 6 megapixels provided (all new images should be taken with this as a minimum resolution).
2. Background of close-ups black.
3. Distance from object to film plane in mm provided (to establish scale).
4. If the photographer is working in a team collecting physical specimens and the photographed plant is actually collected, the catalog number of the physical specimen should be associated with the images.

G. How the collection designations will be used.
1. A user who is just looking for images that illustrate a particular feature in a particular species, or who wants to see a number of images from a particular species would conduct a search of images in the set containing the legacy + regular collection.
2. Users who want to develop resources in which standardization is important (keys, learning tools, etc.), who want to access occurrence records (e.g. GBIF or Flora of the Southeast), or who want to examine the variation in morphology among specimens would restrict their search to the regular collection to ensure that the search results produce the quality of images that their use requires.
3. The point is that by having two tiers of images in the collection we balance the constraints of the photographers (limited time to take many photographs, limited time to go through old images and reconstruct the missing metadata, images already modified in ways that don't meet the standards, preference to take images in ways not specified in the standards) with the demands of users (higher metadata requirements, greater need for standardization of views, higher resolution). When searching the image database, the user will have the choice of more stringency producing fewer hits, or relaxed stringency producing more hits.

Comment summary
A compilation of all of the comments used in the creation of this summary is available at:
http://www.sernec.org/?q=node/214#attachments

Steve Baskauf - Vanderbilt Dept. of Biological Sciences
steve.baskauf@vanderbilt.edu

SERNEC States

LIKE TO JOIN SERNEC?

State-by-State Listing

HERBARIUM SURVEY RESPONDENTS

SERNEC Information

SERNEC Links

The SERNEC Newsletter

Reports and Proposals

Databasing Information

Syndicate