The techniques used previously for compound image compression follow a top-down method that first segments the innovative image into three separate layers, the background layer including smooth objects, the foreground layer including information concerning text and graphics colors and mask layer through using the MRC model. These different layers are compressed independently by a variety of encoders such as JPEG2000, H.264/AVC, and Digipaper algorithm to compress the scanned document. Other presented system uses a block based categorization method that separates pictorial macro-blocks information of text and image. Every of these two types of macro blocks are further encoded by the account of the Multidimensional Multiscale Parser (MMP) algorithm particularly calculated to encode it. The text or non-smooth blocks are compressed by using an MMP - text algorithm.The smooth or image blocks are compressed by using of MMP-Flexible Partition (FP). Since high computational complexity is the main drawback in existing systems and also the high redundancy. Another drawback of scanned images is the poor retrieved image qualities when compressing original images. The proposed Modified MMP Compression algorithms efficiently improve the quality of scanned images and improve the compression ratio and improve the quality of the image. © Research India Publications.