基于红外图像识别的电气设备温升检测

为了更好地利用红外热成像技术对电气设备故障进行识别和诊断的问题,提出了一种基于红外图像特征和种子区域生长法的设备温升检测方法。采用邻域平均法减少了红外图像的噪声干扰,提取出了红外图像RGB空间中的红色分量图及绿色分量图,应用种子区域生长法分别对两分量图进行了分割,先通过寻找局部最高温点划分区域,再计算各区域内形态学梯度筛选出有故障的高温点,并将该点所在区域作为种子区域从而实现种子区域的自动选取,将像素点4个方向上的最大梯度及像素点与种子点的灰度差作为种子生长的判定条件,用交集的方法将分割后的红色分量图和绿色分量图融合,提取出了设备温升过高区域。实验及研究结果表明,该方法能确定高温升区域,且轮廓清晰,为电气设备温升故障诊断提供依据。

Temperature rising detection of electrical equipment based on infrared image recognition

In order to detect the faihJre of electrical equipment by infrared thermal imaging, a new temperature rising detection method was proposed based on IR image features and seeded region growing. The neighborhood average method was adopted to reduce the IR image noise. Red component were green component are selected to make segmentation using seeded region growing. Local high temperature points and morphological gradient were combined to achieve auto seeds selection and tile growth criteria was based on the pixels＇ largest gradient of four directions and gray difference between initial seed and pixels. High temperature regions were extracted by fusing the segmented compo- nent images with intersection method. Experiment resuhs indicate that the method can accurately recognize the high temperature area, and the outline is clear, thus provide the basis for faultal diagnosis of electrical equipment.