基于改进Sobel算子的红外图像边缘提取算法

Edge extraction algorithm of infrared thermal image based on improved sobel operator

针对红外热像仪采集的图像边缘信息模糊,图像显示多样性,边缘信息难提取的特点,提出了一种基于Sobel算子梯度相乘的边缘提取算法。该算法首先对红外热像仪图像进行待识别目标的高温区域提取,然后分别利用增加了6个方向模板的Sobel算子和Roberts算子对图像进行边缘提取,再将得到的两幅梯度幅值图像进行梯度相乘,最终得到边缘提取图像。最后,用MATLAB对图像进行了仿真,仿真结果表明,该算法能够快速有效地提取红外热像仪图像的边缘,弥补Sobel算子的不足及提高了Sobel算子边缘检测的性能,计算简单,具有良好的检测精度,而且得到的边缘较细,极大的改善了图像边缘提取的效果。

The infrared image often has the drawbacks of fuzzy edge, the image diversity and difficulty to extract the edge information. To solve the problems, an edge extraction algorithm based on Sobel operator gradient multiplication is proposed. Firstly, the high temperature region of the target to be identified is extracted ; secondly, Sobel operator added six orientation templates and Roberts operator are used to extract image edge; thirdly, two gradient magnitude images are multiplied; fourthly, edge extracted image is obtained. MATLAB is used to simulate the image. The simulation results show that the algorithm can rapidly and efficiently extract the edge of infrared thermal image, make up the deficiencies of Sobel operator and improve the performance of Sobel operator. The algorithm has simple calculation, high precision and sharp edge. It is effective for infrared image edge extraction.