Other Free Encyclopedias » Online Encyclopedia » Encyclopedia - Featured Articles » Contributed Topics from P-T

Photo Defect Detection

ink intensity detected pixels

Definition: Image inpainting techniques include photo defect detection, where ink traces, scratch, and damage from ink pens are automatically detected.

The concept of image inpainting is discussed in the article entitled “Digital Inpainting.” Most existing inpainting mechanisms allow users to select a defect area or a mask of object to be removed. To cope with the inconveniency, we developed a naive photo defect detectionmechanism. Ink traces, scratch, and damage from ink pens are automatically detected. Weconsider the intensity and shape of photo defects. Though, it is almost not possible todiscriminate ink regions from objects in photos. The intensity variation of ink regions isusually quite smooth and steady while comparing with objects in photos. Thus, we use the HSI color space and use intensity in the first filter. In the second filter, we record the numberof pixels been detected in each intensity variation. We calculate the variance of pixel numbersdetected between two different continuous steps. If the variance is low, it means that pixelsdetected in the last adjustment are not much affected by the present adjustment of intensity. And the pixels been detected are possible to be pixels of ink spray because of its steadyintensity. Keeping on the calculation of variance of pixels detected, a collection consists ofconsecutive adjustments of intensity can be constructed. The collection needs to be analyzedsince sometimes the variance of pixels been detected is low only because it has lowdiscrimination in a step of adjustment. If so, the collection been record has no use and wecan’t detect the ink. Thus, if the collection of adjustment contains too many passes ofadjustment, we just leave it and go on to analysis the photo image. After the calculation andanalysis, a set of adjustment of intensity can be found out to separate pixels of defects fromobjects in the photo. Figure 1 and 2 illustrates our results.

Photogrammetry [next]

User Comments

Your email address will be altered so spam harvesting bots can't read it easily.
Hide my email completely instead?

Cancel or