Individual Weight Estimation of Cynoglossus-gracilis Based on Measurement of Irregular Image Area

Abstract

For aquaculture, the classification of live fish and the deconcentrition of fish into more ponds are very important process. Non-contact and no-harm are always the target of this process. Using the technology of digital image processing, the binary irregular image of Cynoglossus Gracilis is got and measurement; we analyze the relationship between the weight of fish and flat surface area by a linear regression model. The experimental results show that two parameters has a certain linear relationship, the correlation coefficient is 0.3807(R 2 =0.1450). Further research is needed to account for the effects of different growth stages of Cynoglossus gracilis body thickness and other factors, a more accurate model to identify the relationship between the two parameters, to provide support for the non-contact measurement of weight and classification.

Posture-invariant hybrid scaling weight measurement algorithm for live eels

“…Projected area of fish head, fish body, and fish tail Quality detection model of weight and projected area [12] Body length Correlation function of body length and weight [13] Length and width of fish Estimation models for length and width and weight [14] Area of the image Correspondence between image area and body weight [15] Chicken Image projected area, eccentricity, perimeter, target volume, length value, width value, back width Broiler quality model [16] Image surface area, perimeter (the number of pixels in the boundary point) Surface area and perimeter as a function of weight [17] Pig Rear projection area Estimated model between back projected area and mass [18] Shadow area Body mass prediction model [19,20] Egg Perimeter, area, size axis, shape index, and shape factor Mathematical model of the relationship between egg weight and area [21] Mango Horizontal projection area(total pixels) The relationship between fruit weight and its projected image [22] Melon Contour, principal component Single melon weight estimation regression model [23] Consistent with previous work, based on the developed live eel sorting machine, this study presented a method for obtaining the weight of live eels using images. The method did not require complex electromechanical weight detection devices, effectively reduced the damage rate of live eels, and met the practical needs of factory production.…”

“…Li et al [13] applied the computer vision technology in fish grading and sorting, used crucian carp as the research object, and found that the correlation between the power function of fish length and weight was high through image processing and calculation, and proposed a classification method to calculate weight using body length to achieve grading of crucian carp. Viazzi et al [14] used 2D computer vision technology by extracting the length and width of the jade perch in the image establishing shape and group mass estimation for this fish. Ma et al [15] used machine vision techniques to obtain images of narrow-bodied tongue sole individuals, measured the area of irregular images by digital image processing techniques, and used linear regression techniques to establish the correspondence between the image area and body weight, and the correlation coefficient between projected area and weight was 0.3807 as indicated by data analysis.…”