The invention discloses an image block compressed sensing reconstruction method based on a wavelet packet threshold value, which comprises the following steps of segmenting an image according to blocks, converting each segmented block into a column to form a new image, and carrying out compressed iteration reconstruction on the image by adopting a method of minimizing two normal forms on the new image, and carrying out wavelet packet decomposition on an iteration reconstruction result of each time by adopting a wavelet packet. By decomposing an original signal into a high frequency part and alow frequency part, generating a threshold value by utilizing low frequency information, carrying out threshold value processing on an iteration reconstruction result, and generating a control factorby utilizing high frequency information to control the threshold value, and through multiple iterations, a final reconstruction result is obtained. According to the method, the reconstruction result can be well improved.