Detecting Diseases in Jasmine Plants Using Proposed Image Pre-processing Algorithm

It is common knowledge that India’s economy depends heavily on agriculture. Agriculture productivity is the source of economic growth in India. However, plant diseases can directly affect the yields in agriculture and timely detection of diseases is a most difficult challenge for researchers due to the lack of specific techniques in detecting diseases of plants. Since the Jasmine crop is one of the fastest-growing crops in Tamil Nadu state India. This study examines the use of an Image Processing-based Machine Learning (ML) technique, which has five main phases, to identify diseases in jasmine plants. Image acquisition, input image pre-processing, segmentation of the pre-processed images, feature extraction from the segmented images, and disease classification are the phases. Generally, Jasmine plants are affected by many diseases caused by bacterial, viral, and fungi. Among many diseases in Jasmine plants, this research focuses on detecting four types of common diseases, namely leaf spot, rust, powdery mildew, and turning yellow that occur on Jasmine plants. However pre-processing tasks can affect the performance of classification of diseases, there is a need for an efficient pre-processing technique which leads to better accuracy in detecting Jasmine plant diseases. Hence it is significant that the image data is pre-processed before the segmentation task for cleaning to overcome issues related to image data. And to avoid these issues, this paper focuses on image resizing, changing the color space of the images, etc. The image acquisition phase is carried out in real-time image sets which are collected from different Jasmine plant cultivation. Input images of Jasmine leaf with diseases are captured from the normal Android mobile camera or Android tab. The median filter is applied to the input images followed by bilateral filtering is implemented for contrast enhancement. The pre-processed images are taken to the segmentation and feature extraction process for identifying diseases in Jasmine plants. The objective is to develop effective preprocessing techniques for getting better accuracy in detecting diseases of Jasmine plants. This article proposes a hybrid pre-processing methodology for reducing noise in the image and for resizing an image.KeywordsImage processingMachine learningJasmine plantDiseasesMedian filterPre-processing