基于视觉图像的植物受旱等级鉴定研究与仿真

摘要

Research the accurate identification of plant drought levels based on image recognition. This paper put forward a pixel local sensitive discriminate analysis algorithm to complete the identification of plant drought levels. A mapping was used to make the boundary of different plants in the neighborhood maximize. After a mapping, slight differences of pixels in a subspace make drought plant leaves gathered together. The experimental results show that the improved algorithm can improve the identification accuracy.%研究图像识别植物受旱等级准确鉴定问题,一些耐旱植物在吸收不同水分的条件下,叶子的叶片细胞液浓度、比鲜叶重等像素、图像形态特征变化极小.传统的利用图像的植物受旱等级鉴定方法多是基于叶片细胞液浓度、比鲜叶重等像素特征在不同干旱条件下,表现的差异进行植物干旱程度判断,一旦特征差异不明显,其特征变化参数将不准确,造成受旱等级鉴定准确度不高.提出像素局部敏感判别分析算法的植物受旱等级鉴定方法.利用一个映射,使得邻域内不同植物受旱像素点之间的边界最大化.细微差异像素点经过计算映射后在子空间内使得干旱植物叶片点更聚集,克服传统方法的弊端.实验结果表明,利用改进算法进行受旱等级鉴定,提高了鉴定的准确度.