Algorithm for an Automated Clarias gariepinus Fecundity Estimation Technique Using Spline Interpolation and Gaussian Quadrature

Abstract

Fecundity is essential in the field of population ecology, where the number of eggs is measured to get the actual reproductive rate of an organism. The estimation of fecundity is essential for an accurate study of biology and population dynamics of fish species. This estimation can be developed using the gravimetric method (weight method) to calculate the number of eggs. However, this method still requires experienced technicians and much time and effort to compute the number of eggs manually. The increasing growth in both hardware and software have led to many improvements in imaging technology. Hence, this research addresses the problem of employing constructing a computer vision algorithm. This paper introduced the automatic fecundity estimation method, which applied simple mathematic theories and image processing algorithm to estimate the fecundity of African catfish ( Clarias gariepinus ). From the image of the fish, the fish’s eye was be detected using a modified Haar Cascade Classifier Algorithm and appointed axis line where the eye becomes the origin point. Next, we identify the region of interest, which reflects the fish's fecundity to obtain the pixels corresponding to the silhouette of the region as coordinates in Euclidean space, which are then represented with a function using cubic interpolation function. Using this function, we compute the region of interest using an integral numerical approach, e.g., Gaussian Quadrature. From the result, we compared with the ground truth to get the estimation of the number of eggs.