Improvement And Research On PCNN Image Compression Coding

Image compression coding method has developed into second generation. The classic compression coding method based on spatial and frequency of image has fully used the statistical and frequency characteristic, the development potential has been run out. At present, the lossy compression coding method combined with human visual property is the hot point of image compression coding. Pulse coupled neural network (PCNN) is a manual model neural network that derive from the conductive property of the nerve cell of mammal visual cortex. Integrate with human visual property in this paper, apply PCNN into the domain of image compression coding, put forward a PCNN quantizing method that reserve contour information. Then combine with lossless statistical coding method to encode images. Finally, a new improved PCNN image compression coding method is proposed.The main working parts are as follows:1) According to the Machband effect of human vision, utilizing the excellent image segmentation property of the simplified PCNN model SPCNN, extract edge of the image and reserve it. As to rest area, this paper applies the former PCNN quantizing method to process. The ultimate result combines both parts.2) Compare the improved PCNN quantizing method with former PCNN quantizing method on grayscale images separately. Then unite with statistical coding methods to encode images. Discuss the compression ratio variation of the processed images by both methods.3) Apply PCNN quantizing method for RGB colorful image processing. Separate a colorful image into red, green and blue segmental grayscale images. Then use PCNN quantizing method respectively. Afterwards, combine the three gray images into a new color image. For HSV color image format, this literature only needs to quantify the value component, then discuss the quantizing result.4) Using the hardware development platform produced by Timll Corp. and software development kit DVSDK provided by TI to build development environment. After run the test program successful, try to transplant PCNN image compression coding algorithm into the development platform.The experimental results illustrate that the improved PCNN compression coding method has better coding efficiency, the improved method do not has relatively high luminance in grayscale images, and it can highlight the edge data properly at the same time. The PCNN coding method can attain better performance in coding efficiency in colorful image coding. However, the universality of PCNN colorful image compression coding method needs to be enhanced.