【作者】 刘保玲；

【导师】 邱白晶；

【作者基本信息】 江苏大学 ， 农业机械化工程， 2005， 硕士

【摘要】 潜土逆转旋耕耕耘方式是指刀轴位于地表以下逆转旋耕方式。对潜土逆转旋耕耕耘方式众多学者对旋耕刀刀型与切削力和功耗之间的关系进行了系列研究,但由于试验条件的限制,有关潜土逆转旋耕抛土运动的研究还只限于被抛土粒沿旋耕机运动方向的前向运动和后向运动,忽略了土粒的侧向运动。现已取得的研究成果不能反映潜土逆转旋耕的三维抛土特性,但三维抛土特性的研究对于深刻认识潜土逆转旋耕壅土现象,并进而找到减少壅土的有效措施以及提高覆盖质量等问题极为重要。本文在进行了大量的单目高速摄像试验和试验分析的基础上,首次采用高速双目立体视觉系统研究被抛土粒的三维空间运动,建立了提取被抛土粒流序列图像中特征目标的图像处理方法。 根据高速摄像系统的特性、旋耕机具的运动参数及光照条件等确定了在本课题试验条件下的最佳摄像参数,并找到了适用于拍摄被抛土粒运动现象的示踪标记方式。利用摄像系统对多工况下的抛土运动过程进行了拍摄记录,图像分析结果表明:采用示踪标记不仅有助于目视判别,而且可简化图像处理过程,提高目标判别精度。针对示踪标记具有的颜色特性,采用了提取彩色图像中的单色分量,并在此基础上提出采用人工识别和基于小区域的序列图像处理方法来提取运动目标特征,完成了在二维序列图像中对特征目标的跟踪。 在完成单目序列图像跟踪识别目标的基础上,搭建了由MotionPro高速摄像系统和CR2000高速摄像系统构成的高速双目立体视觉系统,建立了基于BP神经网络实施被抛土粒三维运动轨迹恢复的方法。结果表明:在被跟踪土块的运动轨迹是连续的条件下,利用这一方法进行土粒的三维运动轨迹恢复,可保证生成的土粒三维运动轨迹的连续性。 在对潜土逆转旋耕被抛土粒流图像研究中,发现在刀工作的一个周期内,被抛土粒流整体形状的外围轮廓从类似椭圆逐步演化为半圆状。针对这一抛土现象,本文给出了图像几何分区统计描述法。本文还研究了被抛土粒流抛起高度与旋耕机具运动参数之间的关系,发现在刀轴转速相同的情况下,机组前进速度对被抛土粒流起抛高度的影响并不是单调的。更多还原

【Abstract】 Submerged reverse-rotary tillage is a type of tillage with the rotary shaft underlying the soil surface and rotation direction reverse to the tractor wheels’ rotation direction. A series of researches on the relationship between blade type , cut-force and energy consumption have been studied. Due to many factors , all of the researches were focus on the plane motion of the thrown clod and ignored the motion in the side direction, which cann’t reflect the 3D throwing properties of the submerged reverse-rotary tillage. But 3D motion of the thrown clods is most important for researching on avoiding the piled-up clods and improving the cover quality. In the thesis, We developed a method to construct the binocular vision system to research on the 3D motion of clod and developed image processing method to acquire the object characters from the time-arrayed images.According to the properties of the high-speed vidicon, rotary tiller movement parameters and light etc., we found the best parameters of the vidicon and the tracer marker is fit for observing the clod. Image processing results showed that the tracer marker is good for observing the objects in the thrown clods, simplifying the image processing and improving the precision. According to the color properties of the tracer marker, we developed the method of acquiring the component image from RGB image. A method was brought forward to acquire the moving object in the time-arrayed images based on image processing of the region containing objects.Based on the 2D time-arrayed image processing for acquiring objects, We construct the binocular vision system with Motionpro 10000 and CR2000 High speed imaging system. And a method of restore the 3D motion tracks of the clod based on BP Neural network theory were brought forward. The results showed that when the object motion tracking is continuous ,method we developed, can restore the 3D motion tracks with good quality.During our research on the motion rule of the thrown clods, we found a phenomenon of ’similar to ellipse’. In a working cycle of the blade, the periphery of the thrown clods at first looks like an ellipse and then turned to half-arc. In this thesis we give a method named ’image geometry division’ to describe the phenomenon.The relationship between the height of the thrown clods and rotary tiller movement parameters was also researched . The results showed that the moving speed and the thrown soil height is not monotony更多还原