Abstract:A new wood defect detection method based on quick affnity propagation clustering is analyzed.First,the wood image is described by a set of extracted color moment features.In order to further reduce the dimension of the feature set and the distance matrix S,the mean square recovery value is used as diagnostic threshold.Furthermore,the location of the wood defect is automatically recognized and marked.The experimental results show that the proposed method is better than the traditional AP method in terms of computational effciency.The average computing time,accuracy and recall ratio of the proposed strategy attain around 0.557s,70.5% and 95.6%,respectively.

Wood Defect Recognition Based on an Advanced AP Clustering