摘要：One online monitoring method based on the component contents of image feature retrieval is proposed to conduct the extracting and separating process for the rare-earth ion’s feature color. By analyzing different component contents of rare-earth ion color features, it is exposed with the relevance between component contents and the color feature H component. By combining the image retrieval technique based on contents, two retrieval algorithms, H component histogram and H component coherence vector, are proposed to help with the rare-earth component contents’ testing. In order to improve the soft sensing accuracy, the sensing accuracy correction algorithm based on PSO is suggested. To verify the effects of the above algorithm, the rare-earth component contents soft sensing experimenting platform based on image retrieval has been developed with VC6.0 and SQL Server. After doing imitation experiment with the component contents from the subjects of the rare-earth’s two components Pr/Nd extraction separation system, this method is proved with a fairly high testing accuracy.
