Based on compressed sensing theory

we research a meticulous sparsity adaptive matching pursuit algorithm

and propose a new method for digital image retrieval on this basis.Firstly

the original signal of color and vein are formed from RGB color and gray level co-occurrence matrix by order of column prior.Then

these two signals are measured by the blocked compressive sensing method

and measurement vectors are obtained which representing the color and texture features.Secondly

we reconstruct the image by blocks using the MSAMP(Meticulous Sparsity Adaptive Matching Pursuit) algorithm

and calculate the difference and sparse value between the original blocked signals.Finally

we calculate the overall image similarity

and focus on estimating the sparseness of measurement difference.Because it does no need to recover the original signal precisely

so it can reduce the number of iteration and accelerate the retrieval speed.Simulation results show that the retrieval speed and retrieval precision about this image retrieval algorithm based on compressive sensing signal have higher performance.