Video encoder software architecture for VLIW cores incorporating inter prediction and intra prediction

• Main Authors: JAIN ANURAG MITHALAL, SWAMI PRAMOD KUMAR, GOTTARDO DAVID, BRIDE JACQUES MICHEL, MADUMBU VENUGOPALA KRISHNA
• Format: Patent
• Language: English
• Published: 03.07.2012
• Subjects: CALCULATING; CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION ANDCOMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION ANDCOMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THIR OWNENERGY USE; COMPUTING; COUNTING; ELECTRIC COMMUNICATION TECHNIQUE; ELECTRIC DIGITAL DATA PROCESSING; ELECTRICITY; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; PHYSICS; PICTORIAL COMMUNICATION, e.g. TELEVISION; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS; TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINSTCLIMATE CHANGE

This invention is a method of video encoding. The number N macroblocks stored in a temporary buffer depends upon an estimated number of motion vectors. N macroblocks of current and prior frame data is transferred to the temporary buffer. The invention determines for each macroblock whether to be inter frame predicted or intra frame predicted. The inter and intra macroblocks are separately encoded based upon this determination and stored in an output buffer. Output macroblocks are output from the output buffer in raster scan order. This technique permits the process to loop over differing number of macroblocks in differing parts of the encoding process. Entropy encoding complexity from separating inter and intra macroblock encoding is avoided by separating a symbol generation phase from an encoding phase.