FLUDI: Using digital images for measuring fish length

Abstract

Fish length is very important to the fishery research to identify fish population. In many fisheries, sample of fish length is a main parameter to identify fish reproduction, recruitment, growth and mortality. Currently in Malaysia, researchers have to measure the fish length one by one using measurement tools. This method is highly cost as we require buying fish first for measuring purposed. Moreover, manually measurement of fish length is time-consuming. Therefore, a method to measure the length of a fish automatically using image processing and formula optic is greatly needed to solve these prob lems. In this paper, we will discuss on the details of FLUDI framework. The FLUDI framework is a combination between Hsiu's method and Serkan's method. The accuracy of measuring the fish length was 99.81%. Keywords Fish Length; Hsiu's Method; Digital Images IntroductionFishery is a dynamic process and a continuous moni toring and assessment of the population is a need to maintain the sustainable yield of the resources. In fisheries, fish population length data play a major role in determining the population parameters such as age or cohort, growth parameters, mortality rates and recruitment patterns of fish. These biological parameters are useful in predicting the sustainable yield of the fishery resources.The most common method for measuring fish length is the measuring board where the scale is attach to the hard wood or acrylic plastic. The individual fish were measure by using the board with the eyes of the observers should be at the perpendicular to the tip of the measurement. The lengths were recorded in a data sheet. It is time-consuming and labor intensive and further more it is too costly. As the fish resources is a dynamic process which depend on the exploitation rate of current situation, the continuous data collection, monitoring and management from time to time is needed. Failure to do this will create a danger to the depletion of the fish population. As the sampling of fish length data are so tedious, time consuming and costly, we aim to develop an easier method to solve the problem of sampling the fish length for population study.From the previous works, there are quite a number of papers discussing on the methods to measure various objects from digital image (Hsiu 2008;Lee et al. 2004;Naiberg and Little 1994). Naiberg and Little (1994) have developed a size assessment system underwater using model-based recognition and stereoscopic vision. Modelbased recognition is used to locate object and stereo vision system to determine distance and sizes given stereo video input. However, the stereo vision system is expensive, the

Sustainable aquaculture development: a review on the roles of cloud computing, internet of things and artificial intelligence (CIA)

“…2010; Mustafa et al . 2013). The difference in accuracy could primarily be due to the differences in methodologies and algorithms used.…”

“…Interestingly, the use of CIA has shown a promising future in this aspect. Reports indicated that, the accuracy of measuring fish length by processing an underwater fish image ranges from 95% to 99.81% (Rahim et al 2010;Mustafa et al 2013). The difference in accuracy could primarily be due to the differences in methodologies and algorithms used.…”

Overview of Smart Aquaculture System: Focusing on Applications of Machine Learning and Computer Vision

“…The result showed that this is the efficient method to sort fish online and can be applied widely in many fish farms. Mustafa et al [86] proposed this study to measure the length of fish automatically by image processing (FLUDI framework) instead of by measurement tools which are of high cost. Rastrelliger kanagurta and Selar crumenophthalmus species were used for this experiment.…”

“…However, the software can be applied to other objects without any object reference. The actual fish length from the digital image was tested by comparing the fish length obtained from the traditional measuring board (Figure 12) [86]. The result shows that the difference is very small, ranging from 0.74-2.19%.…”