摘要
边缘提取是数字图像处理、条码识别的关键,但是在边缘提取过程中,易受噪声侵染,导致精准度降低,并造成特征信息丢失和模糊边缘。基于此,提出了一种基于霍夫变换联合决策树的边缘提取模型。首先,对图像输入信号进行霍夫转换,降低信号噪声侵染率,缩短线段间隔距离。其次,对像素梯度、像素邻域进行差值分析,并赋予相应权重。最后,利用决策树法搜寻图像特征边缘信号,输出最终图像边缘结果。研究结果显示,霍夫变换联合决策树法能快速有效地提取图像边缘信息,噪声侵染率小于10%、提取准确率和完整率大于90%,且优于标准决策树法。因此,霍夫变换联合决策树能满足图像边缘提取要求,适合用于数字图像处理和融合。
Edge extraction is the key to digital image processing and barcode recognition, but in the process of edge extraction, it is easily infected by noise, which reduces the extraction accuracy, and causes the loss of feature information and blurred edges. Based on this,this paper proposes an edge extraction model based on Hough transform joint decision tree. First, the image input signal is converted to reduce the signal noise infection rate and shorten the line segment separation distance. Then, the difference between the pixel gradient and pixel neighborhood is analyzed and the corresponding weights are assigned. Finally, the decision tree method is used to search for image feature edge signals, and the final image edge results are output. The results show that the Hough transform joint decision tree method can quickly and effectively extract image edge information, and the noise infestation rate is less than 10%, the extraction accuracy and integrity rate are greater than 90%, which is better than the standard decision tree method. Therefore, the Hough transform joint decision tree can meet the requirements of image edge extraction, which is suitable for digital image processing and fusion.
作者
区英杰
OU Yingjie(Guangzhou Embedded Machine Technology Co.,Ltd.,Guangzhou Guangdong 510663,China)
出处
《信息与电脑》
2022年第21期187-189,193,共4页
Information & Computer
关键词
边缘提取
霍夫变换
决策树
图像处理
edge extraction
Hough transform
decision tree
image processing
