MODELING THE DISTRIBUTION OF FISH SCHOOLS IN THE BERING SEA: MORPHOLOGICAL SCHOOL IDENTIFICATION
In this paper we demonstrate how fish schools can be automatically identified from acoustic backscatter data using a simple morphological and threshold algorithm. The algorithm includes: 1) seabed identification and removal; 2) bubble layer removal; 3) choosing a threshold for fish backscatter and s...
| Published in: | Natural Resource Modeling |
|---|---|
| Main Authors: | , , , |
| Format: | Article in Journal/Newspaper |
| Language: | English |
| Published: |
Wiley
1994
|
| Subjects: | |
| Online Access: | http://dx.doi.org/10.1111/j.1939-7445.1994.tb00180.x https://api.wiley.com/onlinelibrary/tdm/v1/articles/10.1111%2Fj.1939-7445.1994.tb00180.x https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1939-7445.1994.tb00180.x |
| Summary: | In this paper we demonstrate how fish schools can be automatically identified from acoustic backscatter data using a simple morphological and threshold algorithm. The algorithm includes: 1) seabed identification and removal; 2) bubble layer removal; 3) choosing a threshold for fish backscatter and setting all pixels below it to 0; and 4) identifying the school boundaries and smoothing the images by using morphological operations. We compared the school sizes and boundaries found by using our algorithms with those found by experts to evaluate our choice of threshold, structuring element and morphological operations. Results showed two classes of images. The algorithm performed well for images having distinct, well defined fish schools. Images having more amorphous aggregations showed significant disparity between the algorithm and experts' choice. Differences appear to involve separating higher intensity aggregations from surrounding lower intensity aggregations. In these cases, the algorithm provides a more consistent and arguably more accurate estimate than the experts. The shoal image data is translated, via a connected‐surface‐identifier algorithm into a table of information about shoal size, shape and location that allows further analysis of the shoal information with a much smaller data set than the original acoustic images. Such analysis can lead to observing patterns in fish shoal distribution and its relationship to environmental factors and to point‐process tests for randomness at different spatial scales which, in turn, allows stochastic modeling of the schooling process. These topics will be presented in subsequent papers. |
|---|