The signal acquisition and compression can be made simultaneously and the signal sampling rate is much lower than the Nyquist frequency in the compressive sensing theory, which provides the possibility for high resolution imaging from low resolution sampling data. A compressive imaging method based on CCD image sensor is proposed. By using the unrepeatable characteristic of the serial output analog pixel value of the CCD image sensor in a single measurement, the semi cyclic and semi random measurement matrices are constructed to compress and measure the analog values outputted by CCD image sensor. Then the total variational algorithm based on augmented Lagrangian (TVAL3) arithmetic is used to decompress and reconstruct the image. This imaging method is good at measuring the sparsity of the matrices, and the original images can be well recovered. The proposed method can greatly alleviate the burden of analog digital and the complexity of quantization coding, which also has a simple structure and strong practicability. Simulation results show that the reconstructed image has better subjective and objective quality with the proposed method.