An Enhanced Feature Extraction Model for the Information Retrieval In Crop Disease Detection

Abstract

Fast similarity search has more importance in a large-scale dataset while using for image indexing and retrieval in many applications. Semantic hashing is an efficient way to accelerate similarity search, which designs compact binary codes for a large number of images so that semantically similar images are mapped to close codes. Among various hashing approaches, spectral hashing has shown better performance by learning the binary codes with a spectral graph. And Color is one of the most important and widely used in content analysis and retrieval. However, most promising color descriptors consume massive amounts of computation and storage, which is a serious drawback. One of these promising color techniques in image retrieval is the Color Correlogram (CC), but the technique also suffers from the aforementioned drawbacks. Compact and conceptual Correlogram descriptor can be used to represent color Correlogram. Compact Correlogram uses compact-generalized Correlogram, which compresses color and generalizes the distances of the actual Correlogram descriptor and conceptual Correlogram uses spatial correlations for the dominant color of few images instead of a large number of quantized color used by the original descriptor. These two representations are integrated with multiple instance learning method to indexing and retrieval of images in both text application and in multimedia applications. Multiple-instance learning has been widely used in image indexing for its capability of exploring region-level visual information of images. Semantic hashing with multiple instance learning leads better results compared to other methodology in terms of indexed precision with respect to image pixels and image bits.