Learn more about this course.

Necessity of data compression

What parameters are involved in compression? In this article, Dr Ming Yan discusses his recent research.

With the rapid development of information processing technology and sensor devices, the spatial resolution, temporal resolution and quantization depth of images are constantly improved, and the amount of data is also increasing exponentially.

In 1972, the first international multispectral imager was successfully launched on Landsat-1 satellite, which only had four spectral bands. The AVIRIS ‘97 hyperspectral data from NASA’s Jet Propulsion Laboratory has 224 bands per scene and each image is about 140MB. China’s “detailed inspection satellite” high-resolution TDI-CCD camera image data acquisition speed can reach 2500Mbps, the US reconnaissance satellite KH-11 is as high as 18Gbps.

At present, the data transmission rate on the satellite is generally 300~600Mbps, while the rate of image data acquired by the charge coupled device (CCD) camera is as high as several or even dozens of Gbps. Therefore, the data transmission capacity on the satellite is far from meeting the requirements of real-time image data transmission. Regardless of the improvement speed of storage and transmission equipment or the socio-economic cost, it is difficult to keep up with the current demand by relying only on the improvement of hardware. The better solution is to compress and encode the image and video data before storage and transmission. Its purpose is to transform and recombine the image and video data according to some rules, so as to represent the information of the source as much as possible with as little data as possible.