The Abnormities Extraction Method of Uneven Illumination Images for Devices Detections

For the abnormity detection with the optical images, the images obtained from an optical system had noise area and uneven illumination background. A combined processing method based on the feature of illumination distribution was proposed in the paper. The method combined open operation and difference of Gaussian filter (DOG) to extract abnormity. By open operation, the background was extracted and abnormities distribution of the dark area was acquired. By DOG, influence of the light source noise was eliminated. The experimental results showed that comparing to the single method, the proposed method let the abnormities distribution of the targets inside the optical images be better emerge. It proved that the proposed method was effective in obtaining the exact abnormities distribution from images taken in the condition of uneven illumination.