摘要
电力监控图像常存在复杂的背景和环境干扰,如杂乱的线路、树木、建筑物等。这些干扰物体可能与目标具有相似的外观特征,使得目标识别难度较大。为此,提出了基于特征阈值分割的电力监控图像目标盲检测方法。计算单一灰度级对应的图像像素,选取直方图区域斜分的阈值,根据灰度级与邻域平均灰度级构建二值图像函数,划分噪声点区和待检测图像点区。基于此,用背景区域分布描述特征一致性,获取分割阈值。通过均值漂移法归一化处理样本数据,利用最近邻搜索方法滤除包含噪声点的冗余特征点。基于超平面法向量与偏差的最大超平面,构建判定目标函数,实现图像目标盲检测。由实验结果可知,该方法最大检测完整程度为99.8%,能够保证检测电力监控图像的完整性。
Power monitoring images often have complex backgrounds and environmental disturbances,such as chaotic lines,trees,buildings,etc.These interfering objects may have similar appearance features to the target,making target recognition difficult.Therefore,a blind detection method for power monitoring image targets based on feature threshold segmentation is proposed.Calculate the image pixels corresponding to a single gray level,select the threshold for oblique segmentation of the histogram area,construct a binary image function based on the gray level and the average gray level of the neighborhood,and divide the noise point area and the image point area to be detected.Based on this,the background region distribution is used to describe feature consistency and obtain segmentation thresholds.Normalize the sample data using the mean shift method,and filter out redundant feature points containing noisy points using the nearest neighbor search method.Based on the hyperplane normal vector and the maximum hyperplane deviation,a judgment objective function is constructed to achieve blind detection of image targets.According to the experimental results,the maximum detection completeness of this method is 99.8%,which can ensure the integrity of the detected power monitoring image.
作者
杨庭
陈闻
严俊
YANG Ting;CHEN Wen;YAN Jun(Hubei Central China Technology Development of Electric Power Co.,Ltd.,Wuhan 430000,China)
出处
《电子设计工程》
2025年第6期118-121,126,共5页
Electronic Design Engineering
基金
湖北华中电力科技开发有限责任公司第一批科研项目(0322HBDL202107114)。
关键词
特征阈值分割
电力监控图像
目标盲检测
均值漂移法
feature threshold segmentation
power monitoring images
blind target detection
mean shift method
