SHRIMP BREED IDENTIFICATION AND COUNTING DEVICE USING IMAGE TECHNOLOGY ON EMBEDDED SYSTEM

Abstract

Abstract: This paper presents the design a study of a device for detect and count shrimp larvae model
using image processing techniques on embedded system that can be applied in industrial sosiety. Prosess
of detecting shrimps, sorting garbage and counting is automatically perform based on hardware,
harware control device and shrimp larvae recognition software. Used mechanical process combine
with photographic chamber to move shrimp larvae and take a image. The shrimp larvae image will
be sent into software in KIT Raspberry Pi 2 for further processing. The main algorithm used that is
the segmentation algorithm based on the optimal threshold and a number of other algorithms such as
background substraction, morphologies,… to aid identification number of shrimp larvae.
Experiments have been conducted on various samples of whiteleg (Penaeus vannamei) larval shrimp. In
each sample, there are a certain amount from 100 to 900 of shrimp larvae that age from 7 to 12 days. The
test results showed that the model worked exactly for samples of up to 500 shrimp larvae that age from 7
to 12 days, with the largest observed error of 2,025%. With a cycle time of approximately 10,34 seconds,
the model can reach 139.265 shrimp larvae per hour.
Keyword: Shrimp larvae, image processing, background subtraction, model of counting shrimp larvae