A Survey on Matching and Retrieval of Near Duplicate Images

The image can be modified using some transformation on original images which forms near duplicate images. An image can be represented by its size or length and the number of patches in the image varies with respect to the length. The adjacent and visually similar pixels are clustered and then the resulting cluster is considered as a patch. Image representation and image similarity measurement are two major issues in image matching. The proposed method extracts patches from given image and represents by variable length signature. The signature is further validated in a near duplicate natural image detection which makes a decision about whether two images are near duplicates or not. The near duplicate image retrieval aims at retrieving relevant images from image database which are similar to query image. The similarity between two images is calculated based on variable length signature to improve the effectiveness.