Вы находитесь на странице: 1из 1

- image partitioning - image is divided into 8 x 8 blocks - why is it divided?

to make resolution invariant - representative color selection - for each block, choose its representative color. We can decide the color by t aking the most dominant color in the region - after that, we transform the color in RGB into Y cB cR space . y = .... cb = ... cr = .. -these will be transformed using discrete cosine transform and then we will obt ain 3 DCT matrices, one for each color component. dct = ... - A sequence 12 number of CLD can be formed by taking first 6 coefficient from Y-DCT matrix, first 3 coefficients from Cb-DCT and first 3 coefficient from Cr-D CT matrices.

The steps to create CLD from an image are as the following: a. An input image is divided into 8 x 8 blocks. Dividing an image into 64 b locks would make the feature invariant to image size or resolution b. For each image block, its representative color is chosen. We can decide the color by taking the average color in the region. After this stage, we have a tiny image containing 8x8 pixels, where each pixel is representative color of i ts region c. Then, each representative color in RGB space will be transformed into Y cB cR space, using this following formula: y = .... cb = ... cr = .. d. -these will be transformed using discrete cosine transform and then we will obtain 3 DCT matrices, one for each color component. dct = ... e. A sequence 12 number of CLD can be formed by taking first 6 coefficient from Y-DCT matrix, first 3 coefficients from Cb-DCT and first 3 coefficient fro m Cr-DCT matrices.

Вам также может понравиться