Title :
Autoscaling software JPEG decoder for embedded systems
Author :
Milinkovic, Stevan
Author_Institution :
Sch. of Comput., Union Univ., Belgrade, Serbia
Abstract :
This paper presents an autoscaling software JPEG decoder with very small memory footprint. An efficient Huffman table processing and fast IDCT calculations make it suitable for embedded systems. Decoding is achieved by processing one MCU at a time and storing it immediately in frame buffer, instead of processing full image strip. There are three modes of downscaling, where 1/8 is extremely fast since it completely bypasses time-consuming decoding operations. Decoder is implemented on microcontroller ARM Cortex M4 using some SIMD instructions for fixed point operations.
Keywords :
Huffman codes; decoding; discrete cosine transforms; embedded systems; image coding; inverse transforms; microcontrollers; parallel processing; Huffman table processing; SIMD instructions; autoscaling software JPEG decoder; decoding operations; embedded systems; fast IDCT calculations; fixed point operations; frame buffer; image strip processing; memory footprint; microcontroller ARM Cortex M4; Hardware; data compression; digital images; embedded software;
Conference_Titel :
Embedded Computing (MECO), 2013 2nd Mediterranean Conference on
Conference_Location :
Budva
DOI :
10.1109/MECO.2013.6601329
