Different from 1D signals

2D images have obvious texture information.Based on the basic framework of block-based compressed sensing

we propose a block-based adaptive compressed sensing of image using texture information

which exploits the property that different texture would produce varied information content.The adaptability lies in two aspects:one is the adaptive sampling and another is the adaptive threshold.Two different filters are introduced

which forms two kinds of new algorithms

respectively.The proposed algorithms are applied to nature and medical test images.The results show that our improved algorithms provide superior performance on both the reconstructed image quality and the visual effect.