Abstract. The management of fish stocks in Chile is in charge of the Undersecretariat of Fisheries and Aquaculture (SUBPESCA). The stocks are subject to catch quotas, which are granted by the Ministry of Economy, Development and Tourism through SUBPESCA. The National Fisheries and Aquaculture Service (SERNAPESCA), supervises, controls and registers that these quotas are exploited by artisanal fishermen registered in the Artisanal Fishing Register.
Nowadays, the control of capture quotas is in charge of a small number of inspection officers and the capture sampling process is not automated or uses any type of technology. This causes two major problems for the control: only a small fraction of all landings of artisanal fisheries can be inspected in the field by the authority and the process of sampling in the field presents diverse sources of error. As a result of this, SERNAPESCA in its fifth axis of the 2015-2018 Strategic Plan seeks to "strengthen the development of information technologies and systems" by strengthening "the oversight model, providing it with capabilities to optimize oversight processes"
The automatic estimation of the total amount landed, by small pelagic species exploited, and by each artisanal fisherman of the country, is a scientific-technological problem of multidisciplinary research and development, since it requires to design: a vision system tuned spectrally to discriminate, in an image, the different pelagic species of interest, and a series of image spectral processing algorithms to calculate the catch percentages of each species. This requires a multidisciplinary expertise of researchers in radiometry, science and imaging technologies and natural sciences.
The conceptual idea to investigate is the reflection that each pelagic species experiences in multispectral light, and it is hypothesized that each species reflects its own spectral trace. It is also hypothesized that some morphological structures of each species are unique, whether in their texture, contour, proportions, among other differentiating characteristics to be identified. Given that each landing groups together fish caught in different areas and at different times (and therefore with different death datas), it will also be evaluated if these two hypotheses are dependent on the death data and the size of each species, which makes the problem of spectral-morphological classification is multidisciplinary. Once the spectral trace of each species has been found, a spectrally tuned vision system will be developed to generate images in the distinctive spectral bands. In addition, this vision system will have classification algorithms that discriminate the type of species observed and, assuming that the total fish landed is weighed, the total amount per small pelagic species exploited by each artisanal fisherman in the country can be estimated. Additionally, exploiting the unique morphological characteristics of each species is intended to develop algorithms that estimate the size distribution of each species of interest.
Publications.
Funding agency: FONDEF.
Program: FONDEF IDeA IV Concurso.
Grant number: ID17I10184.
Funding period: December 2017 — December 2019.
Director: Jorge E. Pezoa.