- In the process of image mining for robot vision, concept analysis is an important technique. The paper proposes a novel framework of image mining for robot vision based on concept lattice theory and cloud model theory. Concept lattice reflects the process of human's concept formation with mathematical formal language. Cloud model is a transformation model between qualitative concepts and quantitative numerical values. Image mining for robot vision is considered as a process of concept extraction from different granularities (image pixels, image pixel groups, image features, image objects, image files and image databases). The methods of image mining from image features(texture features, color features, shape features, spatial relationship features) are introduced in the paper, which include the following basic steps: firstly pre-process images, secondly use cloud model to extract concepts, lastly use concept lattice to extract a series of image knowledge(association rules, clustering rules and classification rules). At last, a software prototype is designed and developed, and some experiments confirm the validity of the proposed framework.
[1]
Tao Cheng,et al.
KDG - Knowledge Discovery from GIS: propositions on the use of KDD in an Intelligent GIS
,
1994
.
[2]
Ji Zhang,et al.
Image Mining: Issues, Frameworks and Techniques
,
2001,
MDM/KDD.
[3]
Zhao Liang.
Research on knowledge discovery based on knowledgebase
,
2008
.
[4]
Deyi Li,et al.
Artificial Intelligence with Uncertainty
,
2004,
CIT.
[5]
Sara J. Graves,et al.
Using Association Rules as Texture Features
,
2001,
IEEE Trans. Pattern Anal. Mach. Intell..
[6]
Bernhard Ganter,et al.
Formal Concept Analysis: Mathematical Foundations
,
1998
.