A novel method for random RS image simulation was proposed and realized. By integrating PROSAIL model, random classes of ground object and scale extension, random point image and random block image were simulated, respectively. In the experiment, several images, including HJ-1 CCD near infrared red spectral band image and the images with different translation, rotation and scale parameters, were quantitatively simulated. With the help of these simulated images, the effect to NIR image registration from the differences of translation, rotation and scale parameters was discussed. The action of block effect in the NIR image registration was analyzed through simulated random block images with scale diversifications and random distribution of ground object. Simulating random remote sensing images were used to analyze NIR image registration. The experimental results show that the block effect in image takes an important role to image registration.