Investigation the Influence of DCT Coefficients on the Image Retrieval Performance
Autor: | Ting Ting Shao, Xiu Ping Zheng, Shui Li Zhang, Jun Tang Dong |
---|---|
Rok vydání: | 2013 |
Předmět: |
Color image
business.industry Trellis quantization ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Pattern recognition General Medicine Euclidean distance Distance measurement Feature (computer vision) Position (vector) Discrete cosine transform Computer vision Artificial intelligence business Image retrieval Mathematics |
Zdroj: | Applied Mechanics and Materials. :1165-1169 |
ISSN: | 1662-7482 |
Popis: | According to the important position of DCT in image coding, the research of image retrieval performance based on DCT is advanced. In the paper, we take color descriptors of MPEG-7 as the color image feature, and acquire new image features through giving different weights to the different elected DCT coefficients, Using Euclidean distance as the similar distance measurement to retrieve color images in DCT. Then the relationship between the choice of weight and retrieval performance is studied by experiments, and moreover the retrieval conclusion is analyzed in detail. The result shows it can acquire relatively satisfactory outcome when selecting right weights of the elected DCT coefficients based on a concrete condition. |
Databáze: | OpenAIRE |
Externí odkaz: |