A Study of Classification of Leaf Shape Categorization of two Stages Based on Fusion | Original Article

ABSTRACT:

The paper work only considered frontal and fresh leaves when creating leaf image databases. In future, CAP-LR can consider and analyze leaves that are wrinkled, occult and dry. Another difficult field of research can be considered in similar ways is color discoloration or discolored leaves. Advanced operations such as parallel workflow processing can be studied to increase the speed of the recognition of the leaf disorder for plant identification. Parallel task processing can be used to group algorithms together during the different recognition stages. This study proposed techniques for improving the operation of plant identification leaf recognition. Positive results from the different experiments show that the model proposals discriminate effectively against the various leaves and identify the right plant to match the image of the inserted leaf. The botanists can therefore safely use this to increase their efficiency in plant recognition and thus save valuable plants in order to improve the quality of human life and life on Earth.A system for improving the leaf image was proposed, called 'Enhanced wavelet-based demoizing with built-in edge enhancement and automatic contrast adjustment algorithm.The 197 method combines the wavelets, CLAHE (contrast adjustment), corner enhances and a relaxed middle filter (noise removal), with a single procedure to increase the visual quality of the leaf image. Texture-based color segmentation technique called 'Enhanced wavelet-based segmentation using the WCF method to extract the leaf image from its background. Five types of features are extracted during extraction geometrical, texture, colour, fractal and leaf-like. These functions are combined to form the GLFS (Geometrically + Leaf), CLFS (Color + Leaf), TLFS (Texture + Leaf) and FLFS (Fractals + Leaf) functionalities. In addition to shared and merged operators to select optimal feature sets two selection algorithms for feature, the genetic algorithm, and the Kernel main component analysis algorithm have been coupled.