弱可视环境下像素级多传感器图像自动分割方法研究

Weak Visual Environment Pixel Level Multi-sensor Image Automatic Segmentation Method Research

像素级多传感器图像分割容易受到外界环境的干扰,在弱可视环境下将丢失大量分割图像的边缘信息,导致分割效果差、分割效率低。为此,提出一种新的弱可视环境下像素级多传感器图像自动分割方法,通过Retinex算法对弱可视环境下像素级多传感器图像进行增强处理。介绍了PCNN图像分割方法的基本原理,通过PCNN方法对输入原始图像进行迭代,对输出结果和输入图像进行"与"运算,获取PCNN一次迭代的像素级多传感器图像,求出迭代后图像的最小方差比,计算出该次最小方差比与上次迭代最小方差比的差值;依据该差值判断是否继续迭代,将最终的输出结果看作像素级多传感器图像的最佳分割结果。实验结果表明,所提方法具有很高的分割效率,且分割效果佳。

Pixel level multi-sensor image segmentation is vulnerable to the interference of the external environment ,weak in visual environment will lose a lot of segmentation image edge information, lead to poor segmentation effect, segmentation efficiency is low. For this, a new weak visual environment of pixel level multi-sensor image au-tomatic segmentation method was put forward, through the Retinex algorithm of pixel level multi-sensor image weak visual environment enhanced processing. The basic principle of PCNN image segmentation method was introduced, through the PCNN iteration method to input the original image, on the output and input image “and” operation, get one PCNN iteration of pixel level multi-sensor image, the minimum variance ratio of image after iteration, calculate the minimum variance ratio and the time the last iterative least variance ratio difference, according to the difference judgment whether to continue the iteration, the final output result as best of pixel level multi-sensor image segmentation result. The experimental results show that the proposed method has high efficiency, segmentation and segmentation effect.