Double Quantization analysis detects the traces left by consecutive JPEG compressions on an image. When a spliced region from one image is inserted into another, if the compression histories of the two images differ, the discrepancy may be detected by this algorithm. A typical case of forgery that is detectable by this algorithm is when an item is taken from an image of high quality (or an uncompressed image, or an image that had its past JPEG traces destroyed by scaling/filtering) and placed in an image of lower quality. If the resulting spliced image is then saved as at a high quality, this should result in a successful detection. In the output map, red values (=1) correspond to high probability of a single compression for the corresponding block, while low values (=0) correspond to low probability of single compression. Localized red areas in an otherwise blue image are very likely to contain splices. Images with non-localized high values and values in the range (0.2-0.8) (green/yellow/orange) should not be taken into account.