The Research Of Heterogeneous PCNN In Image Quantization

In modern society, people get access to peripheral information relying on computer graphics and image processing. With the wave of computer technology, semiconductor technology and mobile internet, the image resolution ratio becomes more and more bigger and so does the amount of information in images. So image quantization, image compressing and transmission technology are all in need of reform and have much room for improvement.Pulse coupled neural networks model inspired by mammalian primary visual cortex neurons is used extensively in the field of image processing, such as image segmentation, target recognition, feature extraction and so on. But it has not been popularly applied to quantization field. So it’s necessary to research into quantization when combined with PCNN and come up with an excellent image quantization scheme that meets the human visual system.The article analyzes the fundamental principles of PCNN and heterogeneous neural network and then proposes Heterogeneous Pulse Coupled Neural Networks algorithm, which are composed of image segmentation and image quantization. The algorithm is validated and tested using Berkeley Segmentation Dataset and standard image library with a good performance. The main work of this paper is as follows:1. Briefly sketch the background and significance of the image quantization research. Summarize the course of development from the aspect of image quantization and neural networks. And then introduce the fundamental principles of PCNN and heterogeneous neural network.2. Introduce traditional image quantization algorithms, such as uniform quantization algorithm, μ-Law and A-Law quantization algorithm, K-means clustering quantization algorithm and DCT quantization algorithm and then make comparisons among them. The experimental results verified that HPCNN algorithm has the best effect on the whole.3. Make use of the hardware platform of ARM S3C2440 to set up a development environment and transplant the HPCNN image quantization algorithm to the development platform which can also satisfy the demand of portability in the Mobile Internet Era.