A PCB Image Self-adaption Threshold Segmentation Method Fusing Color Information and OTSU Theory

Considering that there are many small target pixels on the bare PCB image, the scheme first normalizes the original image in the red, green and blue (RGB) color space, and then converts the RGB color space image to the hue, saturation, and brightness (HSL) color space. Then, skillfully using the intra-class variance and gray difference of unit area ratio of the target and the background as new criterion function to calculate the optimal segmentation threshold based on improved OTSU algorithm in the luminance space. Finally, the segmentation image is obtained with less interfered by noise. The target and background are completely segmented, and the detail information is well preserved, which lay a foundation for the next detection and identification.