Images coded at low bit rates are highly desirable for applications where extensive acquisition, storage, and transmission of images take place. However, JPEG as one of the most popular and widely adopted image compression standard, is known to exhibit blocking artifacts at low bit rates. Our approach to this problem is to develop a subband discrete cosine transform (SBDCT) based coder with the measure of the just noticeable distortion (JND). Our coding scheme offers simple peak signal-to-noise ratio (PSNR) and computation scalability, thus it is referred as scalable perceptual image coder (SPIC). In addition, the compressed bitstream from the proposed scheme is decodable by JPEG decoders; hence permitting easy integration with existing JPEG based systems. Results revealed that the perceived image quality from our proposed scheme is significantly superior to JPEG at the same bit rates.