Image processing apparatus and method that prohibits bi-prediction based on block size

An image processing method, implemented via at least one processor, the method comprising:

• setting identification information for identifying a correspondence relationship between a size of a block of a plurality of blocks expressed by a layer structure of an image and changing of a prediction method applied to the block, wherein the prediction method is changed so that uni-prediction is applied instead of bi-prediction;
• generating a prediction image according to the set identification information;
• coding the image by using the generated prediction image generated, so as to generate a coded stream;
• and
• transmitting the generated coded stream and the set identification information, wherein setting the identification information further comprises prohibiting occurrence of bi-prediction of each block having a size of 8×8 on the basis of a current available memory bandwidth, which is a transmission rate at which a decoded image is read from a memory which stores the decoded image used to generate the prediction image.

An image processing apparatus comprising:

• circuitry configured to
• receive identification information for identifying a correspondence relationship between a size of a block of a plurality of blocks expressed by a layer structure of an image and changing of a prediction method applied to the block, and a coded stream generated by coding the image, wherein the prediction method is changed so that uni-prediction is applied instead of bi-prediction,
• generate a prediction image according to the received identification information;
• and
• decode the received coded stream by using the generated prediction image,
• wherein the identification information is information for prohibiting occurrence of bi-prediction of each block having a size of 8×8 on the basis of a current available memory bandwidth, which is a transmission rate at which a decoded image is read from a memory which stores the decoded image used to generate the prediction image.

An image processing method, implemented via at least one processor, the method comprising:

• receiving identification information for identifying a correspondence relationship between a size of a block of a plurality of blocks expressed by a layer structure of an image and changing of a prediction method applied to the block, and a coded stream generated by coding the image, wherein the prediction method is changed so that uni-prediction is applied instead of bi-prediction;
• generating a prediction image according to the received identification information;
• and
• decoding the received coded stream by using the generated prediction image,
• wherein the identification information is information for prohibiting occurrence of bi-prediction of each block having a size of 8×8 on the basis of a current available memory bandwidth, which is a transmission rate at which a decoded image is read from a memory which stores the decoded image used to generate the prediction image.