基于小波变换和贝叶斯理论的图像分割算法

Algorithm of image segmentation based on wavelet transform and Bayesian theory

基于非下采样小波变换和贝叶斯分类理论的图像分割算法,在处理可见光图像时,容易受到噪声的干扰,而且,原算法在计算最大局部最小点时易产生偏差,从而影响分割效果。因此,本文首先使用滤波器滤除噪声,然后对原来的最大局部最小点的计算方法进行了等价变换,并调整了分割阈值以纠正由于对图像进行小波变换造成的灰度偏移,最后利用数学形态学运算对分割后的图像进行处理以消除孤立点,仿真结果表明改进算法分割效果较好。

The image segmentation algorithm based on undecimated wavelet transform and Bayesian theory is easily interfered by noise and the thresholds are easily calculated inaccurately. Therefore, to overcome the shortage of the original algorithm, this improved algorithm removes noise with a filter, looks for the largest local minimum in an equivalent method, adjusts the threshold to improve the gray shift of the image with wavelet transform, and finally processes the segmented images with opening algorithm. Simulation results show this improved algorithm is better.