Research And Application Of The Segmentation Method For In-field Plant

The weed is the main threat of crops, which led to the decrease of crop yield and quality and the increase of production costs. So the research of automatic identification technology of the weed in the field is quite necessary. The segmentation of plants and background in the field is the basis of automatic identification technology for field weeds. It is an important foundation for the recognition. At the same time, With the rapid development of smartphone upgrading and the popularity of 4G networks, people hope to achieve more function of image processing in mobile terminal. Image processing technology based on Android platform gradually becomes a focus question in the field of graphic image. However, the operation ability of Android platform is relatively limited, which doesn’t meet the requirement of traditional computer for image processing. So, using fast transmission characteristics of 4G networks links the intelligent terminal and the computer to complete complex image processing function.According to the summary and research at home and abroad. the paper gets the plant image in the natural light. and researches preprocessing, segmentation and so on. It provides the feasibility study for automatic identification of the field plant. The main content of the paper is as follows:The image acquired by mobile terminal makes the basic research of image processing with using the method of gray level, threshold segmentation and morphological methods. Finally, the paper makes a summary for the segmentation of field plants and background. Aiming at the shortcoming of basic algorithm effect, the paper puts forward a new segmentation method for field image, based on two-dimensional OTSU segmentation method. And the method brings a neighborhood histogram to reduce the noise and interference of impurity. which makes the number of holds in the image less and refines the edge of plant leaves. The promote of segmentation algorithm lets the next processing result greater and the experiment contrast effects become more obvious.By collating and analyzing the whole process of image processing, it is found that the operation ways of image preprocessing is simple. and fast. And the image after the processing image is gray. which greatly reduces the amount of memory. So the paper proposes to put process of preprocessing to the mobile terminal. and put threshold segmentation, morphology and other methods to the computer. It can not only reduce the consumption of mobile phone flow. but also not affect the processing result. In the end. the UI interface that is a simple operation. clear logical and reasonable is designed in the client terminal. In the server. the full functional and simple interface also is made. The design of both makes the user convenient.