论文部分内容阅读
根据猕猴桃的棚架式栽培方式,提出了一种适用于猕猴桃采摘机器人夜间识别的方法。采用竖直向上获取果实图像的拍摄方式,以果萼为参考点,进行果实的识别,并测试该方法对光照的鲁棒性。试验结果表明:基于果萼能够有效的识别猕猴桃果实,成功率达94.3%;未识别和误识别的果实一般出现在5果及5果以上的簇中,原因是果实相互挤压导致的果萼部分不在果实图像的中心区域,以及果实之间的三角区形成暗色封闭区域;光照过小或过大会导致成像模糊或过曝,对正确率有细微影响;识别速度达到了0.5 s/个。因此,基于果萼的猕猴桃果实夜间识别方法在正确识别率和速度上都有很大提升,更接近实际应用。
According to the scaffolding cultivation method of kiwifruit, a method suitable for night-time identification of kiwifruit picking robots is proposed. The method of obtaining the fruit image vertically and upwardly was adopted. The fruit calyx was taken as the reference point to identify the fruit and the robustness of the method to light was tested. The results showed that the fruit callus could effectively identify the fruit of kiwifruit with a success rate of 94.3%. The unidentified and misidentified fruit generally appeared in the clusters of 5 fruit and 5 fruit because of the fruit calyx Part of the image is not in the center of the fruit area and the triangle between the fruit forming a dark area; light is too small or too large led to imaging fuzzy or over-exposure, the accuracy of a slight impact; recognition rate reached 0.5 s /. Therefore, the fruit calyx-based nighttime identification method of kiwifruit in the correct recognition rate and speed have greatly improved, more practical applications.