Research on microscopic images identification of crack on processed grain fracture surface

To study problems of rice breakage during processing, methods of inspecting rice fracture crack based on image processing were proposed. Rice fracture surface and their microscopic images surface were separately obtained after shear test and microscope scanning. The characteristic parameters of crack were extracted and carried out T-test and BP neural network discrimination. The result shows : the value of parameter K of both kind rice is significantly different. The non-drying and drying rice crack identification ratios were 96.4% and 83.3%.The image recognition method can determine the moisture conditions during rice processing and has a guiding for rice storage and processing conditions control.