STV0674 Functional Description

STV0674

6/35

2

STV0674 Functional Description

The STV0674 uses a combination of hardware functions and firmware to implement the required

features. While the following features are selected and controlled via firmware their operation is

carried out by dedicated hardware core. All dedicated hardware functions use fixed pin numbers

which are detailed in

2.1

Sensor interface

The sensor interface is compatible with ST Microelectronics CIF and VGA sensors. This interface

configuration. All sensor communications, exposure/gain control, color processing, white balance

control, and clocking are handled automatically by STV0674.

2.2

Video processor

The video processor (VP) provides formatted YCbCr 4:2:2-sampled digital video at frame rates up

to 30 frames per second to the video compressor (VC) module or internal video FIFO. The VP also

interfaces directly to the image sensors. The interface to the sensor incorporates:

q

a 5-wire data bus SDATA[4:0] that receives both video data and embedded timing references.

q

a 2-wire serial interface SSDA,SSCL that controls the sensor and the sensor register

configuration.

q

the sensor clock SCLK.

The video processing engine performs the following functions on incoming data

q

full colour restoration at each pixel site from Bayer-patterned input data

q

defect correction

q

matrixing/gain on each colour channel for colour purity

q

auto white balance, exposure and gain control

q

peaking for image clarity

q

gamma correction

q

colour space conversion (including hue and saturation control) from raw RGB to YCbCr[4:2:2].

2.3

Video compressor

The video compression engine performs 3 main functions:

q

Up scaling of input YCbCr 4:2:2 video stream from the VP (typically to scale from QVGA to CIF

image formats),

q

Compression and encoding of YCbCr stream into Motion-JPEG (M-JPEG) format,

q

FIFO monitoring.

The data stream from the VP can be up to VGA size. The scaler in VC can downsize this image.

Once scaled, the video stream is then converted into M-JPEG format. M-JPEG treats video as a

series of JPEG still images. The conversion is released via a sequential DCT (Discrete Cosine

Transform) with Huffman encoding. After transfer through the digiport or over USB, the M-JPEG

stream can be decoded in the host.