Morphology based field rice density detection from rice transplant stage to rice jointing stage

Abstract

ABSTRACT Rice yield estimation is an important aspect in the agriculture research field. For th e rice yield estimation, rice density is one of its useful factors. In this paper, we propose a new method to automatically detect the rice density from the rice transplanting stage to rice jointing stage. It devotes to detect rice planting density by image low-level features of the rice image sequences taken in the fields. Moreover, a rice jointing stage automatic detection method is proposed so as to terminate the rice density detection algorithm. The validities of the proposed rice density detection method and the rice jointing stage automatic detection method are proved in the experiment. Keywords: field rice, density detection, morphology processing, rice jointing stage, plant growing observation 1. INTRODUCTION Rice is one of the primary food crops in the world. Actually, rice yield estimation is an important aspect of the agriculture research field. For the rice yield estimation, rice density is one of its useful factors. In fact, rice density observation is a quite important aspect of the rice growth observation in the field. Moreover, accurate identification of rice density can guide the after agricultural activities in the field. Indeed, Rice density detection is an important content in agricultural meteorological observation. Therefore, rice density detection is a s ubject with significant research value in the agriculture research. However, the density of rice is main ly obtained by observation of the experienced farmers in the past. The manual observation method has many disadvantages. Firstly, manual observation requires a lot of boring fieldwork. Secondly, there is subjectivity in the manual observation result since the experience of the famer can impact the manual observation results. Consequently, different person may get distinct results. Thirdly, the manual data arrangement and data report to upper government is time-consuming and diseconomy. Ahmad et al [1] analyzed the impact of plant density and irrigation regime on grain yield and economic returns with the Cropping System Model (CSM) -CERES. Indeed, they didn’t give a rice plant density detection method. Liu haijuan et al [2] presented an image detection method to detect rice canopy density distribution, and then using the detected canopy density to real-time adjust feeding of the combine harvester, so as to reduce the loss in process of rice harvest. Liu Haijuan’s method only estimated the rice heading density in th e rice mature stage. Therefore, her method still cannot be used to determine the rice planting density. Fumiki Hosoi et al [3] utilized a high-resolution portable scanning lidar together with a lightweight mirror and a voxel-based canopy pr ofiling method to estimate the vertical plant area density (PAD) of a rice canopy at different growth stages. They provided a plant area density (PAD) detection method, but they also didn’t provide a method to estimate the plants number density of the rice. Recent years, electronic technique has been applied to the m odern agriculture research wide ly. Specially, the application of computer vision by the means of the digital image camera is quite popular [4]. To the best of our knowledge, no related literatures about the rice density calculation utilizing com puter vision method have been presented. In this paper computer vision technology is introduced to the topic of rice density detection. The rice image was taken by the color digit camera was fixed in the height of 5m in the rice field. Afterwards, a novel rice density detection method based on the real-time obtained rice images and the corresponding morphological processing is presented. Firstly, the HI-LUT segmentation method is used to segment the rice from the images. Secondly, the morphology processing is applied to the segmented rice image so as to get the rice number in the obse rvation scene. The main idea of the proposed rice density calculation method is to detect the number of rice heaps in the observation scene. Rice heap s in the original rice images are formed by the rice transplantation stage and the distances between rice heaps can be considered as nearly fixed.