Study on leave image processing with application in herbal classification and early detection of chili plant disease

Abstract

Herbs have been widely used in food preparation, medicine and cosmetic industry. Knowing which herbs to be used would be very important in these applications. The current way of identification and determination of the types of herbs however, is still being done manually and prone to human error. Designing a convenient and automatic recognition system of herbs species is essential since this will improve herb species classification efficiency. Chili (Capsicum Annum and Capsicum Frutescen) is an important fruiting vegetable used in majority of Asian dishes. Chili cultivation has been a very difficult and meticulous task due to its vulnerability to various attacks frommicro-organisms, bacterial disease and pests which leave distinguished marks on leaves, stems or fruits. Current manual method applies pesticides and chemicals indiscriminately throughout the farm. To improve the process, development of an automated disease detection is essential. There are a few research that have been done in classification of the plant species using certain factors (leaf shape and size). The classification are accomplished through several image processing techniques. However, the literature shows that there are still a gap in classifying the herb plants species. Therefore, this research focuses on classification approach to the shape, texture features and colors of the herbs leaves. The combination of techniques used in morphology image processing i.e. SVD and skeleton would be able to classify the species of herb regardless of the shape and size. In addition, the techniques demonstrate the capability to detect early plant chili disease through leaf features inspection using HSV colour model technique. The proposed herbs species recognition system employs neural networks algorithm and image processing techniques to perform classification on twenty herbs species. One hundred samples for each species went through the system and the recognition accuracy was at 98.9%. Most importantly the system is capable of identifying the herbs leaves species even though they are dried, wet, torn or deformed. Additionally, a novel method of early automatic recognition for plant chili disease based on color and texture features using a HSV color model and BPNN technique via intelligent decision support system is presented in this research. The proposed system employs image processing technique on one thousand chili plant samples and the recognition accuracy was at 97.7%. The efficiency and effectiveness of the proposed methods in recognizing herbs plant and detecting early plant chili disease are demonstrated by the experiments.