基于计算机视觉的成熟番茄识别研究

Research on Recogniztion of Mature Tomato Based on Computer Vision

以番茄图像为研究对象,提出一种成熟番茄识别方法。首先,以HSI模型中的色调分量为基础进行图像分割,提取出成熟番茄目标图像;然后,再采用最大方差自动取阈值法进行分割处理,对得到的目标图像进行轮廓提取;最后,对轮廓曲线采用Hough变换的方法进行识别,以同一个轮廓圆识别的多个极值点的均值作为最终识别结果,在Hough变换之前采用最小外接矩形法进行有效区域标记,提高了Hough变换的效率。通过多幅番茄果实图像的仿真测试表明:本算法对果实遮掩度为0、小于50%、大于50%这3种情况的识别率分别为78.7%、6 8.1%、4 1.9%,平均识别率达到7 0.6%。本算法对于成熟番茄可以较好识别,尤其对于存在重叠情况的番茄,识别准确率较高。

Takes the image of tomato as the research object and proposes a new kind method of recognizing mature tomato. First,takes the Hue of the HSI model as basis to make image segmentation to extract the image of mature tomato and use the maximum variance automatic threshold to make segmentation. The paper use the Hough transformation to recognize the contour after extract it from the target image and set the mean value of several maximum points of one contour as the value of recognition. Before the Hough transformation it use the minimum bounding rectangle( MBR) to marked the effective region,and this makes the Hough transformation effectively. A plenty of images of tomato was take into simulation test,the algorithm in this paper has the result as follow: 78. 7% with the fruit cover rate 0%,68. 1% with the fruit cover rate less than 50% and 41. 9% with the fruit cover rate more than 50%. The average recognition rate reached 70. 6%,The algorithm proposed in this paper can recognize the mature tomato accuracy,especially for the covered tomatoes,the recognition is accuracy.