Image processing apparatus, data processing apparatus, and image processing method

An image processing apparatus for performing image processing for an input image, comprising:

• N processing modules each configured to refer to a processing result for a reference pixel different from a processing target pixel and generate a processing result for the processing target pixel,
• wherein the N processing modules parallelly generate processing results for processing target pixels different from each other, and
• the N processing modules are connected in a ring shape to transfer the processing results.

An image processing method for performing image processing for an input image, comprising:

• parallelly generating processing results for processing target pixels different from each other by N processing modules, wherein each of the N processing modules is configured to refer to a processing result for a reference pixel different from the processing target pixel and generate the processing result for the processing target pixel, and wherein the N processing modules are connected to transfer the processing results,
• wherein processing results for each of N continuous pixels included in a first pixel line of a processing target region in the input image are sequentially generated in a predetermined order by the different processing module, and
• each of the N processing modules generates a processing result for a first pixel included in the first pixel line and next generates a processing result for a second pixel, wherein the second pixel is included in a second pixel line different from the first pixel line in the processing target region and becomes processable in accordance with the generation of the processing result for the first pixel.

A data processing apparatus for performing processing for a sequentially input data group using a recurrent neural network, comprising:

• a plurality of calculation units each configured to perform an operation corresponding to a hidden layer that forms the recurrent neural network, thereby repetitively generating an output result of the hidden layer and intermediate data to be recursively referred to,
• wherein one calculation unit of the plurality of calculation units is configured to perform the operation corresponding to the hidden layer while recursively referring to the intermediate data for the same hidden layer and transfer the output result of the hidden layer generated by the operation to a different calculation unit, so that the plurality of calculation units perform the processing for the data group using the recurrent neural network.