Progressive Image Transmission Using Edge Detection

In this paper, a progressive image compression and transmission using edge detection scheme is adopted. The image is decomposed into two (primary and secondary) components. Canny method is adopted to detect the edges of the encoded image. These edges are replaced with a pre-designed nine basis nameplates. Then, the Macro edge detection technique is used to reduce the number of these nameplates and keep only the edges that are necessary for visual quality. Eight directional predictions and interpolation technique will be performed on the encoded edges to reconstruct the first layer of the primary component at both receiver and transmitter sides. This is called the 1 stage reconstructed image, which is subtracted from the original to have the 1 stage smooth component. Then, this process will be repeated for the 2 and the 3 stage components. The 3 stage smooth component is filtered using an optimal decomposition filter and then decimated by a factor of 2. The decimated component is encoded using VQ. The decoded result of the smooth image is added to the three layers that form the primary component to have the reconstructed image. An excellent reconstructed images are found at an average of 0.179 bpp (compression ratio 45:1) and with an average PSNR of 39.45 dB. This algorithm is found to be of lower bit rate than existing image compression techniques.