With the rapid development and iteration of CMOS image sensor and other related technologies, in addition to the more advanced technology, the corresponding sensor also integrates the image signal processing system, so it is a leap for the image acquisition system. Based on this, this paper will take the visual intelligent sensor produced and developed by a semiconductor company as the hardware to complete the hardware design of the image acquisition and transmission system. In addition to the core processor of the system, this paper will also use audio and video processing chips and image data processing chips for collaborative analysis, so as to ensure the processing capacity and processing quality of the image acquisition system at the hardware design level. Compared with the traditional system, at the corresponding system software level, in order to further improve the quality of video signal processing, this paper creatively proposes a switch weighted median filter denoising algorithm, which has a significant impact on the processing of random impulse noise in video and the overall image filtering. At the same time, the algorithm also protects the edge details of the video image, thus improving the authenticity of the image and video. Based on this, this paper has built a verification platform based on supporting hardware and corresponding algorithms. Through hardware debugging and system operation analysis, when the operating frequency of the visual intelligent sensor is 100 MHz, the set acquisition depth is 10 bits, and the corresponding system collects data at a rate of 251 fps per second. In terms of the corresponding performance, the iteration number of the algorithm proposed in this paper is about 40% more than that of the traditional algorithm in unit time. In terms of the corresponding algorithm performance, the algorithm proposed in this paper is about 30% higher than the traditional algorithm. In terms of the efficiency of the corresponding image and video acquisition system, the algorithm proposed in this paper is about 44% higher than the traditional algorithm. Based on this, the system proposed in this paper has obvious advantages over the traditional algorithm.
Read full abstract