How does iShield detect explicit images?
iShield’s locates areas of interest in each image using color-analysis. Next a series of techniques are applied, namely contrast normalization, skin-color analysis, texture-analysis, edge-detection, blob-detection, shape analysis, brightness-flow analysis and face detection. The results of these image processing techniques are fed into a neural network to make a decision on the image. Next the text is recovered from the web-page and analyzed. The results of the text-analysis and image-analysis are combined to make an overall decision.