Soybean plant foliar disease detection using image retrieval approaches

• DOI: 10.1007/s11042-016-4191-7
• Publication Date: 2016-12-12T06:04:40Z
• Authors: Sourabh Shrivastava; Satish Kumar Singh; D. S. Hooda
• Keywords: Feature (linguistics); Robustness; Plant disease

A research to solve the soybean foliar disease detection problem is proposed through the image retrieval method. We explore the suitability of image retrieval methods to categorize the diseases of the soybean plants. To solve this problem we present two feature descriptors HIST and WDH, also explore several other colors and texture based feature descriptors such as BIC, CCV, CDH, LBP, SSLBP, LAP and SEH. Research presents an efficient method for soybean disease retrieval and classification method using background subtraction. This research helps to identify the infected lesion pattern which is fully automated and does not require human intervention at any stage. The robustness of the segmentation approach is also verified over six types of the soybean plant diseases. We analyze the statistical and spectral information of the infected portion to model the descriptors. Research test several colors and texture based visual feature descriptors using retrieval based approaches to solve the stated problem.