神经网络与图像压缩编码

Image Coding with Neural Nets

图像压缩编码是一个古老而诱人的领域。而八十年代中期,神经网络的兴起,又为这一领域注入了新的活力。本文介绍了神经网络在图像压缩编码领域中的应用,并提出了一些尚待解决的问题。到目前为止,由于还未寻找到适宜于神经网络应用于压缩编码的好的机理和方法,因此还未获得优于常规编码方法的成果。其中还有一些基本的理论问题(例如,神经网络实现数据压缩的机理)没有得到解决。神经网络是否适于在图像压缩编码领域中使用,以及如何更好地把神经网络用于图像压缩编码领域中去,都是值得进一步深入研究的课题。

Image coding is an old ard attractive area. While in the mid 80's, the prosperity of neural networks gave this area new energy This article gives an introduction to the application of ncural networks to image coding. Up till now, because of lacking method suitable for neural nets to do image coding, neural network image coding method has not achieve better results compared with traditional image coding method like JPEG or MPEG. There are still some fundamental theoretical problems remaining unsolved (for example, the principle of image coding with neural nets). Whether neural networks are suitable for image coding, and how to use neural nets better in this area, is a problem deserves more profound researches.