Image preprocessing for generalized image processing

An integrated circuit (IC), comprising:

• a memory controller configured to access a memory having image data stored therein;
• an image preprocessor, coupled to the memory controller, configured to obtain the image data and generate a plurality of streams of image samples from the image data;
• and
• a processor, coupled to the image preprocessor, configured to process the plurality of streams of image samples;
• wherein the image preprocessor includes:
• a first buffer configured to store a plurality of rows of the image data and output a row of the plurality of rows;
• a second buffer, coupled to the first buffer, including a first plurality of storage locations configured to store a respective plurality of image samples of the row output by the first buffer;
• a plurality of shift registers, wherein each of the plurality of shift registers comprises a second plurality of storage locations comprising an input coupled to outputs of the first plurality of storage locations and an output, the second plurality of storage locations configured to:
• receive, via the input, one or more of the plurality of image samples from the outputs of the first plurality of storage locations;
• store the one or more of the plurality of image samples;
• and
• output, via the output, a respective stream of the plurality of streams of image samples based on the stored one or more of the plurality of image samples;
• an interconnect network including a plurality of connections, the plurality of connections coupling the inputs of the second plurality of storage locations of each of the plurality of shift registers to a different one or more of the outputs of the first plurality of storage locations in the second buffer, one or more of the outputs of the first plurality of storage locations being coupled to more than one of the plurality of connections;
• and
• a control circuit configured to load the plurality of shift registers with the plurality of image samples based on the plurality of connections and shift the plurality of shift registers to output the plurality of streams of image samples.

A method of formatting image data into a plurality of streams of image samples, comprising:

• storing a plurality of rows of the image data, and an output row of the plurality of rows, in a first buffer;
• storing a respective plurality of image samples of the row output by the first buffer in a first plurality of storage locations of a second buffer;
• loading a second plurality of storage locations of each of a plurality of shift registers with the plurality of image samples based on a plurality of connections of an interconnection network, each connection of the plurality of connections coupling an input of the second plurality of storage locations of each of the plurality of shift registers to an output of more than one of the first plurality of storage locations, one or more of the first plurality of storage locations in the second buffer being coupled to a different one or more than one of the plurality of connections;
• and
• shifting the plurality of shift registers to output the plurality of streams of image samples from the second plurality of storage locations.