A Local Threshold Segmentation Method Based on Multi-direction Grayscale Wave

Many disadvantages will impact the threshold segment of a image, such as over light or over dark, uneven illumination and too weak target signal. The paper proposes a novel local threshold segment method based on multi-orient grayscale wave. Firstly it takes the threshold segment in one direction by searching crests and troughs of grayscale wave in one dimension, and then it fuses the segment results in all direction to obtain final segment image. The experimental result indicates that, comparing with global threshold segment algorithms such as Otsu method, minimum error method and maximum entropy method, our method effectively relief the problem of false segmentation and its speed still remains fast.