基于显著性检测与改进Hough变换方法识别未成熟番茄

发布时间：2018-01-04 12:39

本文关键词：基于显著性检测与改进Hough变换方法识别未成熟番茄　出处：《农业工程学报》2016年14期 　论文类型：期刊论文

【摘要】：通过自动识别自然环境下获取果实图像中的未成熟果实,以实现自动化果实估产的目的。该文以番茄为对象,根据视觉显著性的特点,提出了使用基于密集和稀疏重构(dense and sparse reconstruction,DSR)的显著性检测方法检测未成熟番茄果实图像,该方法首先计算密集和稀疏重构误差;其次使用基于上下文的重构误差传播机制平滑重构误差和提亮显著性区域;再通过多尺度重构误差融合与偏目标高斯细化;最后通过贝叶斯算法融合显著图得到DSR显著灰度图。番茄DSR灰度图再经过OTSU算法进行分割和去噪处理,最终使用该文提出的改进随机Hough变换(randomized hough transform,RHT)圆检测方法识别番茄果簇中的单果。结果显示,该文方法对未成熟番茄果实的正确识别率能达到77.6%。同时,该文方法与人工测量的圆心和半径的相关系数也分别达到0.98和0.76,研究结果为估产机器人的多种果实自动化识别提供参考。
[Abstract]:Get the image of the immature fruit fruit through the automatic identification of the natural environment, in order to achieve the purpose of automatic fruit yield. This paper using tomato as the object, according to the characteristics of visual saliency, proposed the use of dense and sparse reconstruction based on (dense and sparse reconstruction, DSR) were detected in immature tomato fruit image, first of all the calculation error of the dense and sparse reconstruction method; secondly based on error smoothing reconstruction error propagation mechanism of context and brighten the salient region; through the reconstruction error of multi-scale fusion and partial target Gauss refinement; finally through the Bayesian fusion algorithm was obtained DSR significant grayscale. Tomato DSR grayscale after OTSU algorithm for segmentation. And the final denoising, using the improved random Hough transform is proposed in this paper (the randomized Hough transform, RHT) circle detection method Tomato fruit clusters of fruit. The results show that the proposed method for proper identification of immature tomato fruit rate of 77.6%. can be achieved at the same time, the correlation coefficient of the radius and center of the proposed method and artificial measurement have reached 0.98 and 0.76, the results of many fruits for estimating robot automatic recognition to provide reference.

【作者单位】：海南师范大学信息科学技术学院;华南农业大学数学与信息学院;
【基金】：海南省高等学校科学研究项目(Hnky2015-24) 国家星火计划(2015GA780002) 广东省自然科学基金(S2013040015381) 海南省自然科学基金(20156225)
【分类号】：TP391.41;S641.2
【正文快照】： 马翠花,张学平,李育涛,林松,肖德琴,张连宽.基于显著性检测与改进Hough变换方法识别未成熟番茄[J].农业工程学报,2016,32(14):219-226.doi:10.11975/j.issn.1002-6819.2016.14.029 http://www.tcsae.orgMa Cuihua,Zhang Xueping,Li Yutao,Lin Song,Xiao Deqin,Zhang Liankuan.

【相似文献】