一种水下图像目标边缘分割方法

A Method to Detect the Edge of Underwater Image

水下图像中目标边缘的检测精度决定了目标的定位和尺寸检测精度,水下目标一般为管状,目标边缘定位难度大,本文针对以上应用环境提出了一种边缘精确定位的方法。首先采用区域分割的方法得到目标的粗略位置,然后细化目标得到管状目标的中轴线,最后用参数拟合的方法来提取目标边缘,既有效抑制了噪声和背景的干扰,又提高了边缘定位的精度。

The location accuracy and size detection accuracy of underwater object is determined by the edge detection accuracy of object; and the underwater targets are generally tubular, so it＇s hard to get precise edge position. A method to detect edge position is proposed in this paper. First, the approximate location was obtained by region segmentation;then the central axis of target was determined by image thinning; and at last, the edge was precisely detected based on parameters fitting method. The disturbance of background and other noises are reduced effectively, and the edge position accuracy is improved by this method.