Image processing techniques for internal texture evaluation of French fries

A computer vision technique was used to evaluate internal hollowness, an important textural attribute of French fries. Three computer vision algorithms (co-occurrence matrices, texture features corresponding to visual perception. and run-length methods) were modified, implemented, and evaluated on color French fry cross section images. Evaluation and classification of internal hollowness were based on textural features in two color coordinate systems (RGB and 1(1), 1(2), 1(3)). Statistical procedures were used to determine the classification capabilities of the selected significant texture features of the three algorithms. A maximum of 100% classification accuracy was obtained using any of the three algorithms.