A comprehensive study of feature extraction techniques for plant leaf disease detection

Agriculture has been the most primary source of the livelihood of man for thousands of years. Even today, it provides subsistence to about 50% of the world population. Plant diseases are the serious cause of big losses to crop production every year worldwide. It is necessary to keep the plants healthy at various stages of their growth/development to deal with the financial losses from plant diseases. Symptoms of infections are visible mainly at plant leaves; thus leaves are commonly used to detect and identify the diseases. Detecting the disease through visual observation is itself a challenging task and requires a lot of human expertise. Image processing techniques along with computational intelligence or soft computing techniques can be used to provide a better assistance for disease detection to the farmers. A disease in plants can be detected based on its symptoms extracted in the form of features. Feature extraction techniques thus play a vital role in such systems. The paper emphasizes on the review of hand-crafted and deep learning based feature extraction with their merits and demerits. It provides a comprehensive discussion on a variety of image features such as color, texture, and shape for various disorders in different cultures.