一种基于边缘模式的直方图构造新方法

基于图像边缘和噪音模式的分析，使用了一种目标和背景之间的边界检测方法；并基于边界的描述，提出了一种新的在边界两侧和边界内部选取相等数目的像素构造直方图的方法。该种直方图避免现有方法中全部像素直方图、加权直方图和内部像素直方图不适合于小目标的缺点，避免了边缘像素直方图抗噪能力差和阈值因图像边缘类型型变的缺点。该直方图能同时用于大目标和小目标时以及边界是阶跃边缘和屋顶状边缘时的阈值选取，具有很大的通用性和实用性。实验结果证明，使用该方法的直方图优于现有的直方图构造方法。

NEW HISTOGRAM MODIFICATION BASED ON IMAGE EDGE MODEL

In this paper, a new histogram modification method is introduced, which is based on the image edge model and noise model. Its main consideration is using the image edge model and noise model to detect the boundary between object and background, and then choosing the pixels sited in the left and right side of the boundary, or pixels sited in the boundary, to build the histogram. With this method, in the histogram, the number of pixels in object is approximately equal to the number of pixels in background. This new method can avoid the defects of losing small object in conventional histograms, such as edge intensity weighted histogram, histogram of interior pixels and histogram of whole pixels, and can avoid the defects of sensitive to noise and threshold depending on the model of edge in the histogram of pixels having high edge values. The new histogram is very general and practical, which can be used for the threshold selection when the object is big or small and the edge model is ramp like or roof as well. Experiments prove that this new method is rather better than others. In recently years, little literature is found in the field of histogram modification, and this new method makes a big improvement.