Fish Larvae Counting System Using Image Processing Techniques

Abstract

Abstract This paper presents the use of computer technology based on image processing techniques to count the number of fish larvae with less time processing. Computer technology used is as an alternative solution to the manual counting approach method in term of determination fish larvae survival rate, stock assessment and monitoring fish growth population. Generally, the fish larvae counting is performed with sequential process with laborintensive task which difficult to be used for counting large sample dataset. Traditional counting method has been used for many years, however many researchers highlighted several drawbacks of the manual counting process such as time consuming, laborious, required human skills-eyes, less-accurate, less consistent, difficult to estimate with many large sample and too many involve with human intervention. Since the problems is addressed, many researchers are interested to develop many techniques to facilitate the process of fish counting with fast assessment. In this study, we present combination of image processing techniques that consists of image enhancement, edge detection, and thresholding process. Meanwhile, the blob analysis is used as the statistical information to measure the objects properties in the image automatically. Total number of 150 samples dataset of Nile Tilapia (Oreochromis niloticus) were used in the experiment. All the samples are divided into three part which are small dataset (50 samples), medium dataset (50 samples) and large dataset (50 samples). The performance of the proposed method and the manual approach method are compared based on the number of fish that was successfully estimated and processing time taken through the experiment. All 150 samples of fish larvae were collected from the Freshwater Hatchery of University Malaysia Terengganu. The experimental results shows that the proposed method based on computer technology is outperformed compared to the manual counting approach in the experiment. This is because, the number of fish larvae measurement by using the proposed method is almost similar and some of samples present accurate result compared to the traditional approach. Moreover, the proposed method is promising on the processing time for measuring all samples with less time processing and more reliable.