摘要
研究对图像的树木腐蚀区域分割问题,提高分割区域大小的准确性。传统方法根据图像中物体与背景像素的颜色区别,利用像素视觉差异大小的方法进行木材腐蚀区域图像的分割。但是在木材图像中,腐蚀区域的颜色和树木的本体颜色极其接近,边缘像素的一致性会造成干扰,像素相减的结果会发生错误,造成不能完整的分割木材腐蚀区域的问题。为提高识别准确性,提出利用高斯一维线性模型的树木腐蚀区域分割方法。通过抑制边沿像素的对称性和单调性,去除边沿像素中的干扰因素,克服了传统方法的缺陷。实验证明,改进方法实现了在与自身颜色相似情况下木材腐蚀区域图像的完整分割,取得了满意的效果。
Research segmentation problem of trees corrosion image and the improvement of the accuracy of segmented region size. Traditional method uses the color difference between object pixels and background pixels and the visual difference method to segment the wood corrosion area. But in the lumber images, the colors of corrosion region and the colors of trees wre extremely close, the subtraction operation may results in the error. This paper put forward a linear Gauss model for the segmentation of trees corrosion. By inhibiting symmetry and monotonicity of the edge pixels and removing interference factors frin edge pixels to overcome the defects of the traditional method. Experiments show that the method realizes the complete image segmentation under the case that the wood color is similar to the corrosion area.
出处
《计算机仿真》
CSCD
北大核心
2012年第1期260-263,共4页
Computer Simulation
