机器视觉在幼苗自动移钵作业中的应用

该文提出了一套在自动幼苗移钵作业中用于幼苗生长状况检测的机器视觉系统。穴盘中幼苗的图像被采集和处理,识别出适合进行移钵的单元,用于自动幼苗移钵机的移钵作业。相邻单元的幼苗边缘重叠和叶片挤压会造成识别错误,在该研究中以番茄幼苗作为试验样本,使用基于形态学的分水岭算法处理来完成叶片边缘分割,提取每个穴孔中幼苗的叶片面积和叶片周长来确定适合进行移钵的单元。试验结果表明该机器视觉系统识别准确率达到了98%,应用于自动幼苗移钵机器人中可以很好地判断不同生长状况的秧苗生长质量。     

高山生态有机红茶牛肉干的研制

在传统五香牛肉干制作配方和工艺条件下,采用不同添加方式和添加量将毕节高山生态有机红茶添加到毕节市威宁黄牛肉中,制作高山生态有机红茶风味牛肉干,通过对加工工艺和技术要点进行探讨,分析了影响高山生态有机红茶风味牛肉干制作的相关因素。结果表明,高山生态有机红茶添加到牛肉的方式以红茶抹茶撒拌效果最好,烘烤时喷淋茶汁浓缩汁其次,在复煮时加入茶叶比从初煮时加入茶叶口感和风味评分较高,同时得出各类添加方式的最适红茶添加量。     

草坪草屑的加工与利用

我国每年修剪草坪产生大量的草坪草屑,对草坪草屑的加工利用,可以对我国能源、饲料、肥料及蛋白资源的短缺起到一定的缓解作用,因此如何对草坪草屑进行合理利用已成为一项重要的研究课题。本文综述了草坪草屑能源化、饲料化、肥料化和深加工4种利用方式,详细介绍了草坪草屑在发酵生产能源、饲喂家畜、堆肥还坪以及加工叶蛋白等方面的研究进展,指出禾本科草坪草屑和豆科草坪草屑应采用不同的处理方式,对各加工方式的投入和产出也应进行进一步的量化核算,同时总结并展望了草坪草屑利用方式的发展前景。     

基于局部二值模式和灰度共生矩阵的籽棉杂质分类识别

籽棉杂质的分类识别是实现棉花生产线自适应加工的基础与重要依据。该文提出了一种基于局部二值模式和灰度共生矩阵的籽棉杂质分类识别算法,该算法将含杂籽棉图像首先转换为局部二值模式图像,获取图像的微观结构,再用局部二值模式图像生成灰度共生矩阵并计算特征参数,获取图像宏观结构。使用支持向量机作为分类器,用不同尺度的图像结构进行训练,从而达到籽棉杂质的分类识别。试验结果表明,该文设计算法对各种杂质的平均正确识别率达到了94%,超过单独使用局部二值模式和单独使用灰度共生矩阵的正确识别率,为实现棉花自适应加工提供了技术基础。     

超级杂交水稻理想株型图像识别的探讨

为提供评判超级杂交水稻理想株型的方法和手段，以Visual C 为工具，利用图像识别技术，提供了超级杂交水稻理想株型经图像的二值化、细线化处理、提取了株型的骨架信息。     

基于Xception-CEMs神经网络的植物病害识别

随着深度学习技术与农业的密切融合,越来越多的研究将深度学习技术用于农业病虫害检测,提高农产品产量和质量.本文提出一种新颖的基于Xception模型的植物病害识别方法.了解到植物病害图像会受到不确定环境因素的干扰而减小图像信息.在Xception的基础上,提出一种新的通道扩增模块,采用带有通道分配权重的多尺度深度卷积与组...     

排种器性能检测中种子位置智能检测方法

根据基于计算机视觉的排种器性能检测的特点，提出了基于神经网络的种子位置智能检测方法。采用两个光轴互相垂直的摄像机监测投种规律，并把以矩法求取的质心图像坐标作为输入节点，构造检测种子质心实际坐标的两个神经网络，并行处理左右两幅图像，从而快速求得种子位置。实践表明，此方法不仅能求出种子位置，而且可全面监测种子的运动情况。     

Evaluation of three non‐genetically modified soybean cultivars as ingredients and a yeast‐based additive as a supplement in practical diets for Pacific white shrimp Litopenaeus vannamei

Two growth trials and a physiology assessment were conducted to evaluate three non‐genetically modified (GM) soybean cultivars as ingredients in practical diets for Pacific white shrimp, Litopenaeus vannamei. In addition, a commercially available fermented yeast product was evaluated as dietary supplement. For the growth trials (46 and 35 days, respectively, for trials 1 and 2), the basal diet was primarily composed of soybean meal (SBM), fishmeal (FM), whole wheat, corn protein concentrate, poultry meal (PM, pet food grade) and corn starch. Non‐GM cultivars were processed with novel methodologies to produce Navita? ingredients (N1, N2 and N3) which were incorporated at low (L) or high (H) levels into the experimental diets, in partial replacement of FM and full replacement of conventional SBM. The last two formulations incorporated the fermented yeast for a total of nine experimental diets (Table 1 ). Results from the growth trials indicate that shrimp fed diet 5 (HN2) exhibited significantly lower (p < .05) weight gain as compared to shrimp fed diets 1 and 2 (basal and LN1, respectively) in trial 1, as well as compared to animals fed diets 1, 2, 3 and 8 (basal, LN1, HN1 and basal + yeast) in trial 2. The feed conversion ratio significantly increased for shrimp fed diet 5, in contrast with shrimp fed diets 1, 2, 3 and 8 in trials 1 and 2, as well as compared to shrimp fed diets diet 6 (LN3) in trial 2. For the physiological assessment (stress and immune responses), only the effects of diets 1, 3, 8 and 9 (basal, HN1, basal + yeast and HN1 + yeast, respectively) were investigated. Granular cell counts were significantly higher for shrimp fed the yeast‐containing diets. Haemolymph glucose and haemolymph packed cell volume were significantly reduced for shrimp fed diets 3, 8 and 9. No significant differences were observed in total haemocyte counts, hyaline cells counts, semi‐granular cells counts, haemolymph protein, haemocyte phagocytic capacity and haemocyte respiratory burst activity. Results of this work indicate that selective soy breeding technology coupled with novel processing options has the potential to increase the nutritional value of conventional SBM for shrimp feeds. Trends on immune responses were more difficult to elucidate possibly due to the limited length of the feeding trial.     

蚯蚓处理香蕉茎叶二因素配方筛选

以切碎状香蕉茎叶与牛粪按不同比例混合养殖蚯蚓,探索用蚯蚓处理香蕉茎叶,将香蕉茎叶转化为有机肥料并能最大量获得蚯蚓作为高蛋白动物饲料的二因素处理配方.结果表明:添加80%香蕉茎叶的处理产出蚯蚓数量最多,所产有机肥料含氮最高;添加90%香蕉茎叶的处理虽然产出蚯蚓数量略少,但蚯蚓的重量和80%处理所得蚯蚓重量差异不显著,而且能处理掉更大数量的香蕉茎叶,同时所产有机肥料含有机质更高.     

基于隐马尔科夫模型的深度视频哺乳母猪高危动作识别

哺乳母猪的高危动作和仔猪存活率有密切关系,能直接体现其母性行为能力,而这些高危动作又与其姿态转换的频率、持续时间等密切相关。针对猪舍环境下,环境光线变化、母猪与仔猪黏连、猪体形变等给哺乳母猪姿态转换识别带来的困难。该文以梅花母猪为研究对象,以Kinect2.0采集的深度视频图像为数据源,提出基于Faster R-CNN和隐马尔科夫模型的哺乳母猪姿态转换识别算法,通过FasterR-CNN产生候选区域,并采用维特比算法构建定位管道;利用Otsu分割和形态学处理提取疑似转换片段中母猪躯干部、尾部和身体上下两侧的高度序列,由隐马尔科夫模型识别姿态转换。结果表明,对姿态转换片段识别的精度为93.67%、召回率为87.84%。研究结果可为全天候母猪行为自动识别提供技术参考。