Computer Assisted Photo-Identification - An Overview

The process of computer-aided individual identification of wildlife involves diverse concepts. In general, the following steps are necessary:

  1. Acquisition of photos: This step requires techniques to obtain high-quality digital photos of the species under investigation.
  2. Insertion of the digital photos into a data management system: This step is necessary to associate metadata, such as time and place of encounter, gender, age etc. with each acquired photo.
  3. Preprocessing the photographs: In this step, the quality of the pictures enhanced to be easier usable by the subsequent individual identification process. For example, contrast can be enhanced, specific regions of interest can be cropped, specular reflections can be removed etc.
  4. Identification of individuals: Finally, for each new obtained photo, it needs to be compared to all known patterns in a species' database. Computer-aided photo-identification makes it possible to find an unknown pattern in databases of thousands of pictures. The computer provides a list of the patterns that are most equal to the unknown pattern, and the researcher makes the final decision.

Even though the computer aids in the identification process, the processing of an unknown pattern can be very time-consuming, depending on the executed steps. Assuming thousands of unknown photograps, it is of great importance for the researcher to reduce the time spent for the processing of each photo. On the one hand, an intuitive photo handling reduces the probability of operation error. On the other hand, a shorter processing time allows the researchers to spend more time on field work, data analysis etc..

AmphIdent aids the researcher for individual photo-identification of amphibians by integrating steps 2-4 into a seamless process: Once the photograph has been stored, AmphIdent takes over:

  • AmphIdent allows the researcher to add metadata to the image, such as animal length or gender and age.
  • AmphIdent integrates all required preprocessing of the images. Its image processing algorithms manage to equalize contrast, semi-automatically extract the region of interest, detect important areas in the patterns and automatically perform image enhancement procedures to the images. These integrated features make an additional manual preprocessing needless.
  • AmphIdent automatically suggests the best matching patterns for an unknown pattern. Then, the final decision is left to the researcher. This process is fully integrated with a database, making at possible to automatically update your MS Access database with the newly processed picture. Alternatively, AmphIdent can generate a text-based table file that is readable by MS Access.
  • AmphIdent makes it simple to batch-process all pictures in a given directory. AmphIdent remembers, where you previously left your work and it automatically ensures that no pattern is forgotten or processed twice.

Do, 10 Sep 2015 - Maximilian Matthe