Today’s emerging LED based lights for cars and automatic traffic signs are mostly not continuous light sources but pulse light based. When capturing these light sources with state of the art ADAS sensors, annoying flicker effects can be observed. These flicker effects can be reduced or even fully eliminated with Dream Chip’s sophisticated Deflicker methods.

Dream Chip is using a combination of algorithms. The computational loads of these methods depend on quality requirements. In addition the customer platforms are very different ranging from software based implementation (NVIDIA Tegra K1/X1, Texas Instruments TDA2/TDA3) to hardware based implementations (XILINX Zynq, Altera).

Depending on the customer specifications for implementation platform and Deflicker requirements, Dream Chip will compile solutions from its Deflicker IP portfolio which will perfect fit to customer needs.

Supported Platforms

Software based

• Texas Instruments TDA2/TDA3 family
  • SW for EVE and C6x

•  Nvidia Tegra K1 and X1 devices
  • SW for CUDA and ARM/Neon

Hardware based

• XILINX Zynq 7000 SoC/FPGA devices
• Altera Arria HPS/FPGA devices
• Including SW for local ARM/Neon

Requirements & Features

•  Requirements depend on sensor resolution and implementation type and platform. Typical values (720p60/1080p30) are:
  •  32-64 Mbyte DDR memory space
  • 1-3 Gbytes/s memory bandwidth

• Software based
  • CPU with SIMD extensions
  • Local memory required

• Hardware based
  • App. 60 logic units FPGA space
  • App 64 kbyte block RAM 

• Low latency (2-3 frames)

Home Back