Abstract: Three-dimensional imaging technology is a low-cost, portable and online screening tool for optical non-destructive testing, which allows the obtainment of the space information of samples. Three-dimensional biospeckle was applied in the prediction of beef texture to minimize the detection time and reduce the limitations of its application. The images of samples were acquired at two different angles, pre-proccessed by generalized difference (GD), and fused by gray template matching, wavelet transform, and contrast modulation fusion separately. A temporal history of speckle patterns viz. inertia moment was used as the speckle activity to establish a prediction model for beef texture. The traditional camera calibration method was used and the internal and external parameters of the camera were established. In-depth information from the images was extracted based on the similar triangle principle. The results showed that three-dimensional biospeckle could enable better prediction of beef texture. Among three image fusion methods, wavelet transform provided the best prediction performance, with correlation coefficients of 0.944 4 and 0.928 8 for the prediction of hardness and chewiness, respectively.

Key words: biospeckle, three-dimensional imaging technology, beef, texture characteristics