基于MATLAB的苹果图像分割方法.docxVIP

基于MATLAB的苹果图像分割方法 侯纯利 （曲阜师范大学物理工程学院 曲阜273165） 摘要：本文以苹果图像为例，重点研究了苹果图像与背景的分割，提取出苹果图像，并对苹果的大小进行估算分级，由于苹果品种众多，其颜色，果型，大小均有差异，并非一种或几种算法便可解决苹果分级问题，本文针对市场上最常见的红富士苹果，以背景设置对照物方式，采用Otsu方法进行阈值分割，通过计算苹果的像素点数量与背景对照物像素点数获得苹果上视图面积，再通过多次对比面积与重量，通过对果型的分析，确定苹果上视图面积与重量的简单函数关系。并以此为思路，提出关于水果分割及水果分级的简单过程。 根据实验设计，以matlab为工具，针对分割苹果背景图像时，需要人工设置阈值的问题，重点研究了图像阈值。选择Otsu法，将苹果图像分割为二值图像。采用区域标记与数学形态分去除二值图像中非苹果区域将得到的二值图像与灰度图像进行点乘运算，实现苹果图像与背景的分割，用相同的方法，实现设置参照物与背景的图像分割。该方法简单且能将苹果区域较为完整的从背景区域中分离出来。 仿真结果表明，本文提出的背景分割算法不仅能在红富士苹果图像实现较好的处理，对于其他相似的苹果图像也同样适用，具有一定的实用性。 关键词：水果；图像分割；MATLAB；面积计算 Abstract With the apple image as an example, focus on the apple image and background image segmentation, extraction of apple, and apple on the size of the estimated grade, because the apple variety, its color, fruit type, size difference is not one or several algorithms can solve problems according to the classification of apple, the most common on the market of Fuji red apple, by setting control method, Otsu method using threshold segmentation, through the pixel calculation of the number of apples and background contrast pixels in Apple view area, after many times of size and weight, based on the analysis of fruit types, determine the simple function view area and weight the apple. And as a way of thinking, this is a simple process of fruit segmentation and fruit grading. According to the experimental design, using MATLAB as a tool, the problem of manual setting threshold is needed for the segmentation of apple background image, and the image threshold value is studied. Select Otsu method, the apple image is divided into two value image. The region labeling and morphological remove binary image in non Apple areas will be the second value image and the gray level image of point multiplication and realize the segmentation of apple image and background, using the same method, set up the reference object and background image segmentation. The method is simple and can separate the apple area from the background area. Simula