Skip to Main Content
An innovative embedded algorithm based on context model using neighboring coefficients and parent coefficients is proposed, which can implement real-time image processing and can achieve high fidelity. By analyzing the algorithm and implementation, it is found that the wavelet transform and bit-plane prediction have a great perplexity which can not be implemented efficiently. Based on multiple DSP hardware platforms, a architecture using parallelism and pipelining technology is present, which can reduce computation complexity and improve encoding efficiency. Experiment results show that the algorithm can achieve higher performance than JPEG2000 in terms of image quality and operating speed.