基于直方图修正的小波自适应阈值大米垩白分割

在大米垩白分割过程中,由于噪声等影响,图像原始直方图呈明显的锯齿状变化.在阅值选择定位过程中,会产生大量虚假阈值点,影响阈值点的自适应准确选取.通过直方图修正,实现直方图的平滑处理,消除大量虚假阈值点,保证意义阈值点的位置与值不变,达到阈值选取过程中抗噪声的目的.并利用小波变换实现阈值点的自适应多分辨率定位.实验表明通过直方图修正和小波自适应定位多阈值可实现大米垩白区域的有效分割.

Wavelet automatic thresholds rice chalk segmentation based on histogram adjustment

In the process of rice chalk segmentation, the image histogram is obviously serration changing under the influence of noise and some other factors. There are many false thresholds affecting the threshold points automatic selecting. Hence, the histogram adjustment was used to smooth the histogram, eliminate many false thresholds and keep the real thresholds location and value unchanging. At the same time, the wavelet transform was used to locate the thresholds automatically. Experiments indicate that using histogram adjustment and wavelet transform locating thresholds can segment rice chalk area effectively.