摘要
计算机视觉鸭蛋品质检测中,目标与背景的有效分割尤为重要。为了解决传统灰度阈值分割存在的弊端,构造图像的灰度-梯度共生矩阵,提出基于灰度、梯度信息和最大熵原理的二维阈值分割方法。通过统计目标和背景的熵,并使二者和最大,确定与此对应的灰度、梯度值,即为最佳分割阈值。采用数学形态学方法对分割图像后进行处理,去除噪声点,使分割效果更理想。实验表明,该方法有效。
In duck's egg quality inspection based on computer vision,effective segmentation of target and background is particularly important.In order to solve the deficiency of traditional gray level thresholding and makeup gray level-gradient cooccurrence matrix of the image,two-dimensional thresholding method based on gray-gradient information and maximum entropy principle is proposed.The optimal thresholding segmtentation is that through the statistics of target and background entropy,which makes target and background entropy's and to maximum,the corresponding gray and gradient values determined.Post-processing segmented images by using the method of mathematical morphology,noise points are removed,segmentation is even better,experiments show that the method is effective.
出处
《农机化研究》
北大核心
2011年第8期62-65,共4页
Journal of Agricultural Mechanization Research
基金
淮安市科技支撑计划项目(SN0965)
江苏省高校自然科学研究项目(10KJD520001)
关键词
计算机视觉
鸭蛋
灰度-梯度共生矩阵
二维阈值分割
最大熵
computer vision
duck's egg
gray level-gradient cooccurrence matrix
two-dimensional thresholding segmtentation
maximum entropy