Online Measurement of Content in Block Frozen Shrimp by Fusion of X-ray Imaging and Binocular Vision

Abstract

A machine vision system based on the fusion of X-ray imaging and the binocular stereo vision was developed for the online estimation of net content in block frozen shrimp. Supported algorithms were specifically developed and programmed for the online system, including image acquisition, processing, controlling the whole process, and saving the classification results. The results indicated that the relationship between mean gray value of X-ray image and net content of shrimp is linearity. Meanwhile, the coefficient of linear model also can be represented by the thickness of block frozen shrimp. An online estimation model was built with mean gray value and thickness as dependent variables. Binocular stereo vision technology was also employed to acquire thickness information of samples to revise the estimation model. The performance of the predictive model using two variables was achieved, with R (correlation coefficient) of 0.9475 and root-mean-square error of prediction (RMSEP) of 22.0993 in prediction set. Good consistence confirmed that the proposed method has significant potential application in online estimation of content in block frozen shrimp.