Investigation and implementation of content-based retrieval system for wheat pest images

Taking wheat pest images as a case study, techniques in a content-based image retrieval system were studied including image feature extraction, image similarity measurement and user relevant feedback technologies. Color moments based on four overlapping division and a similarity measurement method based on BP neural network were prompted for the improvement of image retrieval. Besides, the grey relevant feedback was improved to accomplish semantic based image retrieval. A content-based pest retrieval system was developed. And the evaluation results on this system through the sorting evaluation method, precision rate and recall rate show that it is practical to a certain extent, which provides technological support for quick crop pest diagnosis and recognition, and the sharing of crop pest images.