| 摘要 |
PROBLEM TO BE SOLVED: To realize high performance inspection, even if a noise exists by extracting the failure of an object to be inspected based on a noise removing image data of only an image data of an object to be inspected except for a noise area. SOLUTION: An object 4 to be inspected is et on an inspection stage 5 and is photographed by a photographing device 1 for storing the image data of an object to be inspected. A noise area is extracted from the image data to generate a noise area extraction image data. Then, the noise area of the image data of an object to be inspected which the noise area extraction image data show is substituted by a representative value of an image element of the noise area surroundings to obtain the noise area removing image data. Then, in a failure extraction processing, a two-dimensional distribution of a light quantity and a sensitivity at the time of luminance system such as a light diffusion plate 6, an luminance light source 7 and a stabilized power source 8. Furthermore, a smoothing processing, a second differential processing and a re-smoothing processing are carried out to generate a second smoothing noise removing image data. Thereafter, a binary processing is carried out to generate a failure extraction image data. Next, this failure extraction image data is subjected to quality decision for processing quality decision data. The quality decision is made in comparison with a decision criterion separately stipulated.
|
