Retrieval of similar pictures on pictorial databases
References (5)
- et al.
Similarity retrieval of iconic image databases
Pattern Recognition
(1989) - et al.
An intelligent image database system
IEEE Trans. Software Engng
(1988)
Cited by (90)
A new relevance feedback technique for iconic image retrieval based on spatial relationships
2009, Journal of Systems and SoftwareState-of-the-art on spatio-temporal information-based video retrieval
2009, Pattern RecognitionCitation Excerpt :Apart from the 2D string and its variants, some other spatial models are also used in computer vision. These models have gained less popularity but none the less offer a novel and fresh perspective for solving the problem of image and video retrieval, such as 2D projection interval relationships (2D-PIR) [56], the nine direction low triangular (9DLT) matrix [27], qualitative pairwise salient region patch [57], CRTs [26], attributed relational graphs (ARGs) [58–61], θ–R representations (also called spatial orientation graph) [30,62], and mixed directional relations-based quadtree representation [63]. Some studies compare spatial relationships in a weak manner based on spatial partition of the image without image segmentation, such as weighted five fuzzy regions [23], 2D tagging 5×5 grid scheme in CAETI Internet multimedia library [64] and the entities and attributes model [65,66].
Archival image indexing with connectivity features using randomized masks
2008, Applied Soft Computing JournalCitation Excerpt :A data structure called discrimination tree, consisting of a hierarchical arrangement of kd-trees [2] is also proposed for efficient indexing of images. There are two principal approaches to image indexing: (i) those based on the features extracted from raw image data [9,11,13,14,23,26,30], and (ii) those based on the coefficients in the compressed or transformed domain [15,17,20,33,34]. A majority of the techniques proposed in the image indexing literature extract robust features, meant for high similarity matches.
Graph-based representation for similarity retrieval of symbolic images
2008, Data and Knowledge EngineeringCitation Excerpt :Directional similarity [5], a more flexible matching, has been proved to be NP-complete [26]. In addition to these object-matching methods, some algorithms hash or index the extracted features of all database images as a hash table or a tree structure [3,12]. Each feature has a linked list of image identifiers that contain this feature.
Symbolic image indexing and retrieval by spatial similarity: An approach based on B-tree
2008, Pattern RecognitionRetrieval of images by spatial and object similarities
2008, Information Processing and Management