摘要 |
The present invention provides an image processing method applied to a graphics processing unit including: receiving data of an image, wherein the image is a first rectangle constituted by a plurality of pixels, and the pixels of the image are represented by a plurality of image values with a predetermined bit depth in the data; using a predetermined number of bits to perform an accumulation of the image values of the pixels for constructing an integral image of the image, wherein the predetermined number of bits is less than log2(W×H×2k) number of bits, W is the width of the first rectangle, H is the length of the first rectangle, and k is the predetermined bit depth. |
主权项 |
1. An image processing method, applied to a graphics processing unit, comprising:
receiving data of an image, wherein the image is a first rectangle, the first rectangle is constituted by a plurality of pixels, and the pixels of the image are represented by a plurality of image values with a predetermined bit depth in the data; using a predetermined number of bits to perform an accumulation of the image values of the pixels for performing an integration operation on each of the pixels of the image, wherein the predetermined number of bits is less than log2(W×H×2k) number of bits, W is width of the first rectangle, H is length of the first rectangle, and k is the predetermined bit depth; obtaining a plurality of integral values of the pixels by the integration operation, wherein at least one of the integral values is overflow value; and storing the integral values into a volatile memory to construct an integral image of the image. |