Research On Parallel Processing Of Digital Images Based On Multi-core Environment

At present, digital images are increasing in number, the amount of the data needed to be processed is also growing. people have higher requirements for digital image processing technology, the computer needs to process more complex and more images, it means that image processing will consume more time. Digital image parallel processing(DIPP) based on the parallel algorithm and parallel environment is implemented to achieve the rapid disposal capability, the main purpose is to improve the speed and efficiency of image processing.Combining the parallel processing and digital image processing algorithm, using the digital image parallel processing model, parallelize the representative algorithms of the digital image data layer, and on this basis, a double layer parallel color edge extraction algorithm is proposed in this paper, which solves the problem of long execution time caused by serial processing of several algorithms; and a serial algorithm for video object tracking is parallel designed.The main research work and achievements of this paper are summarized as follows:Firstly, this paper reviews the DIPP research background and the present condition, the parallel processing technology based on multi-core environment is studied, including parallel correlation analysis, parallel decomposition strategy, parallel programming model,implementation model, and the evaluation and standard to measure the parallel performance.Secondly, the research contents and characteristics of digital image processing are analyzed theoretically, according to the characteristics and process of digital image processing,the parallel parts are found, then the parallel processing and parallel transplantation model of digital images are studied.Finally, the algorithms for image segmentation, enhancement, restoration, and target tracking is parallel analyzed and designed based on the multi-core environment, through the experiments, the serial and parallel processing time of each algorithm in different size image environment are obtained, the speedup ratio and efficiency of each algorithm can be calculated. Through the performance analysis, the experiment proves the validity of the DIPP under multi-core environment.