Detection and recognition of wood defects based on gray transformation and BP neural network.

Wood defect and rot debase wood quality badly. X-ray as a method of measurement was adopted to detect wood defects nondestructively. Due to the changed intensity of x-ray which crossed the object, defects in wood were detected by the differences of X-ray absorption parameters. Therefore images were processed and analyzed by computer. Gray transformation could enhance the contrast of the image obviously and the position of rot could be highlighted. Binary processing was employed for the image after gray transformation. The defects areas of the binary images were filled. On the basis of image processing of nondestructive testing and characteristic construction, defects mathematic models were established through using characteristic parameters. The feature parameters were preprocessed and were input into BP neural network, and then the wood defects could be recognized. The experimental results show that the detection rate can be up to 90% and the performance shows that this method is very successful for detection and classification of wood defects. This study provides a new method for automatic detection of wood defects. It is useful for the scientific selection and effective utilization of wood resources.