應用於植物工廠作物之生長量測影像系統

Abstract

近年來針對植物進行有效率的自動化量測，使用影像處理技術與雙眼視覺進行非破壞性量測成為了主流的方法。儘管雙眼視覺可計算物體的深度，有量測三維特徵的優勢，但仍存在需要改進的地方，因此本研究嘗試使用近年來受到矚目的時差測距攝影機開發應用於植物工廠的作物生長量測影像系統。 本研究結合了自行開發的重量感測裝置與時差測距深度攝影機於植物生長量測，建構一套輕便型可自動化拍攝的整合移動平台，能夠簡易放置在立體式栽培層架的平面上。透過定時操作馬達移動攝影機，由上而下拍攝種植在層架上的作物生長資訊，包括彩色影像及深度資訊。藉由本研究開發的影像處理技術，系統從不同平行位置拍攝的層架畫面可接合成環場影像方便整個種植平面的觀測，並計算結合深度資訊的幾何特徵，可達到自動化平台影像紀錄、生長曲線計算、重量感測等功能。 本系統經驗證後與使用立體視覺量測技術的效果進行比較，並嘗試使用奈米氣泡對波士頓萵苣進行生長量測實驗。經實驗結果分析，成功建立了影像與重量的生長曲線，奈米氣泡則是具有可提升萵苣重量的效果。

Nowadays the efficient plant growth measurement, using image processing and stereo vision methodology takes the priority as the nondestructive measurement method. Despite the fact that stereo vision can measure spatial information of the plant, it still has some drawbacks. Therefore, in this research, we try to build a plant growth imaging system using the new popular Time-of-Flight camera. In this research, a time-of-flight range camera has been applied to plant growth measurement along with a self-made weight measurement device, to develop a portable, movable and automatic plant growth measuring system. The hardware design enables the system to be mounted on a planting bed of the multi-layer cultivation shelves. And then the system will control the motor to move the camera, taking color images and depth information from top of the view. By using the image processing methodology developed in this research, images from different perspectives can be stitched into a panoramic image, providing a convenient whole view of the planting bed. The system will combine time-of-flight information to calculate plant’s geometry features and the functions like automatic image measuring, growth curve calculation and weight measurement can thus be realized. After the validation, we compare the results done by the stereo vision techniques and a Boston lettuce growth experiment which add nanobubbles has been examined in this research. According to the result, lettuce adding nanobubbles during the growth process can enhance the total weight.