Pemilihan parameter pengolahan citra yang optimal untuk penghitungan benih dan telur ikan

Abstract

The aim of this research is to determine the optimal image processing parameters consisting of thresholding level and ratio of fish school area in order to count the number of fish jouveniles or eggs with image processing method. The thresholding level was determined with two methods: 1) Percentile of fish image grey scale, and 2) Pointing the fish image color. The percentile threshold value for 0.5 cm gouramy was 75, 3-4 cm catfish, 1 cm tilapia, and 1″ pangasius were 70, 5 cm carp and 1 cm gouramy were 50, eggs of gouramy was 90, and 3 cm red tilapia was 25. Ratio area of jouveniles fish school with average area of one jouvenile has different value depent on the fish type. Ratio of fish school eggs was bigger than fish school. Percentile of grey scale threshold showed accuracy for jouveniles of catfish was 90.07%, 1″ pangasius was 84.75%, 1 cm gouramy 86.11%, 0.5 cm gouramy was 86.09%, and eggs of gouramy was 91.72%. Fish image color threshold showed accuracy for jouveniles of tilapia was 93.33%, 1″ pangasius was 90.00%, catfish was 80.65%, 1 cm gouramy was 93.29%, 0.75 cm gouramy was 89.82%, 0.5 cm gouramy was 84.90%, and eggs of gouramy was 91.72%.