الملخص

Classification is widely and largely used in data analysis, and pattern recognition. The data analysis aims to discover similarities between them and group them based on similarity into multiple classes. Artificial intelligence techniques are characterized by their great ability to classify objects and classify images. In this research, some artificial intelligence algorithms, represented by swarm optimization algorithms, were used to detect and classify plant diseases to healthy and unhealthy through images of different leaves of plants. Where plants are considered one of the most important organisms on this planet because of their important and fundamental role in the continuation of life and in achieving environmental balance, as well as in the economic side in many countries, and other benefits of high importance. These plants are apt to many different diseases. As a result of the technological development that the world witnessed in various areas of life, it was necessary to make use of it in the field of plant disease diagnosis, as many artificial intelligence techniques were employed in the discovery and diagnosis of plant diseases. In this paper, a new method is proposed to classify and distinguish a group of eight different plants to healthy and unhealthy based on the leaf images of these plants They are apples, cherries, grapes, peaches, peppers, potatoes, strawberries, and tomatoes using a hybrid optimization algorithm. In the first stage, the plant leaf images were collected and pre-processed to remove noise and improve contrast. In the second stage, the features were extracted based on the statistical feature extraction method, while in the third stage, the particle swarm (PSO) and chicken swarm optimization(CSO) algorithms were used to diagnose and classify plant diseases. Then these two algorithms were combined to produce a proposed hybrid algorithm called (PSO-CSO) hybrid method. The results obtained from these three algorithms were compared and the proposed method (PSO-CSO) obtained the best results compared to the two methods. Where the proposed method obtained in the first and second tests a diagnostic rate of (96.9%) and (98.18%), respectively.