Feature extraction model in systems of diagnostics of plant diseases by the leaf images

Abstract

The task of features extraction occurring in creation of systems to diagnose crop diseases using leaves images has been examined in this work. The model of diagnostic feature extraction of leaves images has been proposed for the problem solution. The main concept of the proposed model includes creation of preferred feature set simplifying the construction process of considered rule in pattern recognition given in the form of images. Experimental investigations at problem solving on diagnosis of wheat diseases with the help of leaves images have been conducted for function test of the presented model. Keywords: diagnostic systems, base fragments of images, diagnostics of plant diseases, diagnostic features of leaves, preferred features, correlation of features.

Algorithms for plant disease diagnostics by leaf image

“…Determination of the difference function 𝑑(𝔏 𝑢 , 𝔏 𝑣 ) between two leaf images 𝔏 𝑢 and 𝔏 𝑣 . At this stage, in the space of preferred diagnostic features (𝔛 ̃𝑗), the difference function that describes the difference between the two, that is 𝔏 𝑢 and 𝔏 𝑣 leaf images, is determined, where 𝜒 ̃𝑗 is the set of preferred features determined by formula (6). When constructing the function 𝑑(𝔏 𝑢 𝔏 𝑣 ), the following principle is used: "the greater the value of the function 𝑑(𝔏, 𝔏 𝑣 ), the greater the difference between these images.…”

“…One of the main tasks of using modern information technologies is related to the creation of computer systems designed to diagnose crop diseases and predict their development. In recent years, a number of research papers devoted to the diagnosis of plant diseases, in particular [1][2][3][4][5][6], have appeared. The problem of creating a computer system based on processing of images of cultivated plant leaves, which is intended for diagnosing diseases of their fruits, is considered.…”

“…The introduction of automated diagnostic systems makes it possible to use unbiased information on the diagnosis of cultivated plants, to diagnose plant diseases in a timely and sufficiently accurate manner. This creates conditions for making decisions on plant protection measures [5,6]. Image recognition methods and algorithms play a key role in the creation of computer systems designed to diagnose plant diseases.…”

“…are more common. When solving the problem of diagnosing plant diseases by leaf images, the processed data are expressed in the form of a numerical matrix [5,6,21,33]. In this case, there is too much information in the leaf image to make a diagnosis.…”