Conventional optical imaging require the image plane at the focal plane of the imaging lens group to get the maximum flux. So the focal length has great impact on imaging quality. In order to investigate the influence of focal length on quality of computational ghost imaging, the experiment of computational ghost imaging was constructed, and the probability density of the measurement of bucket detector was utilized to analyze the results of computational ghost imaging with different focal length and background light. The results show that the variation of focal length and background light have little effect on the quality of computational ghost imaging, which solve the problem that the defocus lead to the deterioration of quality caused by the inequality of two optical path in traditional ghost imaging. Computational ghost imaging can be realized without imaging lens and only using a bucket detector, thus avoiding the allocation of flux on the pixel dimension as well as improving the signal-to-noise ratio. Therefore, computational ghost imaging is very suitable for imaging detection under the extremely weak background.