| 摘要 |
<p>PURPOSE:To reduce the circuit scale for filter processing and to make the circuit arrangement efficient and increase the calculation speed by performing the image filter processing by compressing pixel bits of a stored image. CONSTITUTION:Matraix of 3X3 pixels is extracted from the image stored in a memory 14. Here, the bit width of each pixel is 12-16 bits. As for the matrixes, a pixel bit compressing means 15 which uses a look-up table compresses the bit width from precision of 12-16 bits that each pixel of the matrixes has to precision of 8 bits of the input part of a D/A conversion part 17 and overwrites the resulting matrixes on the previous matrixes. For the overwritten matrixes, an image filter means 21 performs calculation by using a weighting matrix and overwrites the resulting matrixes again to obtain new matrixes. Consequently, the circuit scale is reducible, the circuit arrangement is made efficient, and the calculation speed can be increased.</p> |
