基于边缘信息检测的BP网络图像压缩新算法

A New Algorithm of Image Compress Based on Edge Information Detection BP Neuron

探讨了人工神经网络用于图像压缩的方法,提出了一种基于边缘信息检测的BP神经网络压缩方法:按图像各部分像素特征将图像分为平滑块、纹理块和边缘块,不同的分类块采用不同的隐含层数.实验结果证明:本算法在保证重建图像丰富细节的同时,能大大提高图像的压缩比.

The algorithms of image compressing used artificial neuron are discussed, and a new algorithm of image compress based on edge information detection and BP neuron is developed. This algorithm classifies the image into three kinds of parts according to the pixel character,the smoothness blocks, the texture blocks and the edge blocks. The neuron for different kinds of blocks has different hidden layers. Experiments have demonstrated that the compress ratio is largely improved while preserving the abundant image details.