Image compression is one of the most important branches of digital image processing. It reduces the size of the captured images and minimizes the storage space on the drivers to speed up the transferring and transmission. In this paper we will present a new approach for compressing stereo images based on three algorithms; the first one is comparing the two images that perform the stereoscopic view by noticing the great similarities between them and encoding the difference between the two images instead of encoding the whole image. The second one is reducing the redundancy between the Pixels using a 2D Digital Curvelet Transformation so we can utilize the great ability to represent the curves in the image with minimum number of coefficients. Then quantize them and remove undesirable coefficient. The low number of coefficient contains most of image data. Last one is using Huffman Encoding and take advantage of the lossless property so we can encode image and reduce the size of data without getting any image distortion or lose any part of this image. The performance of the proposed algorithm evaluated using Compression Ratio standard which is the number of the image bits after compression to the number of the original image bits before compression. Also, Peak Signal to Noise Ratio standard (PSNR) which represent the similarity between the restored image and the original image. In final, the Mean Square Error standard (MSE) which represent the error between the restored image and original image. In conclusion, the main objective here is to get the lowest rate for image compression ratio with the highest value for the image quality PSNR at the lowest value of the errors MSE.