N128
Image sensor coupled to a analog neural preprocessing unit. Array of 128x128 McCulloch Pitts neurons with nearest-neighbor coupling, enabling preprocessing like diffusion, binarisation and segmentation.
1-CHIP CMOS-CAMERA
Image sensor coupled to on-chip digital signal processing unit. The image can be preprocessed to automatically correct gamma, fixed-pattern noise and defect pixels.
VISP 128
Visual Instruction Sensor Processor (VISP):
Analog Processor for 8x8 Matrix Multiplications (as in JPEG). Focal plane operation,
demonstrator imager with 64x128 Pixels. Analog pixel data is multiplied with digital
coefficients by digitally controlling the current paths in the analog processor.
A computational ADC performs binary weighted sum in a error-correcting current mode ADC. The analog processor is sufficient for JPEG compression with good quality.
The concept is fast enough to be scaled to 720 lines / 30 fps.
AMS-Neuro
Array of 4096 spiking neurons with nearest-neighbor coupling via adaptive Hebbian synapses. Used for image segmentation, with areas of similar gray scale values resulting in a wave-like synchronization.
Vision_IC Neuro
Mixed-signal processor for 128x128 analog image feature detection. Pixels are processed by a nearest-neighbor integrate-and-fire neural network with adaptive Hebbian synapses. Also used as a test chip for 3D connection technology.
Active-Pixel-Sensor
Active sub-100nm pixel image sensor to test structures in varying depths in the substrate, establishing different spectral sensitivity. This enables color processing without a color mask.
High Speed CMOS Sensor (HSCS)
128*128 pixel 300 FPS high-speed image sensor with digital feature processing (binarization and object tracking). Also used as a test chip for an ADPLL.
Vision_IC CCam
16384 digital pulse computation and routing cells, mimicking neural microcircuits. Pulsing pixels in each cell provide optical input. The overall chip enables neural-microcircuit-like image convolution similar to visual cortex