| The construction of scientific instrument network lab, can resolve the problems exist in scientific instruments, which include three aspects:inadequate instrument resource, unbalanced distribution, and low utilization efficiency. Network lab consists of two main parts:lab comprehensive management, which provides advanced management to guarantee scientific instruments available in 24 hours; and scientific instrument remote experiment, which provides on-site experimental information, for remote users to participate in the experiment through network, and make interactions by the means of text, voice, video, it is convenient for the studies and researches.The monitoring method of scientific instrument work status based on image recognition technology, separating the status monitoring from scientific instrument and controlling computer, with a fully objective and effective recording of instrument work time. It is a critical technique of lab comprehensive management, which provides technical means of scientific instrument management and evaluation. But general image recognition method is designed for natural images. And when it is used to screen image recognition of workstation, the algorithm's computational speed is slow, which is unable to satisfy the needs of real-time status monitoring. So screen image recognition method of scientific instrument workstation is one of the major research content of this paper.Workstation screen sharing can provide instrument running status, analysis testing parameters, operation process, experiment results and other information for remote users participating in the experiment. It is main access to experiment information for the remote users, and also the critical part of remote experiment of scientific instrument. In current screen sharing technology field, natural image coding method is commonly used in image compression. Natural image coding method does not consider the specificity of workstation screen sharing of scientific instrument, and it keeps a large space for improvement in image coding speed and compression ratio, Therefore, according to features of workstation screen image, image compression method is another major research content of this paper.This paper focused on the workstation screen image of scientific instrument to carry out the research on image recognition and compression technology. The research of image matching method and search strategy is carried out to enhance the speed of image matching. The research of image color quantization method and data coding is carried out to improve image coding speed and compression ratio. Specific contents are as follows:(1) A fast image matching method based on projection featureScientific instruments status monitoring, based on recognition technology, involves screen image matching method. The matching method based on projection transformation is adopted, because it is easy to achieve, and has high computing speed, and monitoring precision. But traditional projection matching method has the problems of low design efficiency of template and calculation redundancy in matching process. According to these problems, a fast image matching method based on projection feature is proposed in this paper. Firstly, the calculation workload analysis of projection matching process is carried out. On this basis, target object's characteristics of the minimum aspect ratio rectangle feature is considered to construct a parallelogram template, and description of the boundaries is realized by the use of four-direction chain code. Then in the projection transformation process, gray value real-time determination mechanism is added to reduce the redundant calculation of apparent non-matching points. In this process, dynamic threshold SSDA algorithm is also used to optimize the matching process of one-dimensional array after projection transformation. As a result, image matching speed is improved. Experimental results show that the computing speed of proposed image matching algorithm is 49 times faster than standard projection matching method. As a general image matching algorithm, it can also be used for other image matching applications.(2) Research on the search strategy of workstation screen image featureSearch strategy is a key part of image recognition technology, which determined the comprehensive performance of image recognition method, In common image search strategy, there are some redundant searching spaces, and the relativity of prior results are not used. As a result, search efficiency of this search strategy is low. This paper proposed a kind of image search strategy based on automatic learning-style search model. Firstly, based on the restrict relationship and moving range of testing object, image searching space is narrowed. And then, classification statistics of a certain number of prior points are made to guide next searching process. Experimental results show that, when the proposed method is applied to scientific instruments workstation screen image recognition, searching space can be narrowed by 70%. Furthermore, compared with traverse-type searching strategy and multi-resolve pyramid searching strategy, the proposed method is 1-2 quantitative level higher in the case of same matching method is adopted. Besides, compared with traverse-type searching strategy, genetic algorithm and hill-climbing algorithm, proposed method has advantage in recognition precision and computing speed, and can satisfy the demands of real-time monitoring of scientific instrument status. Meanwhile, this design idea can be applied to other fields.(3) Research on color quantization method of workstation screen imageColor quantization technology is a effective way to minimize image color quantity. Common color quantization method with slow computing speed, is unable to meet the needs of real-time quantitative of screen image. Furthermore it may cause the lose of important detail information of workstation screen image in the color quantization process. For such problems, a color quantization method adapted to workstation screen image is proposed. Firstly, a weighted distance value is introduced to effectively adjust the similarity degree of different colors, so as to decrease the possibility of detail color being combined. Secondly, color quantization process is optimized. The influence with image quantization of similar color of screen image window title bar is removed by pre-process. Meanwhile, for the color quantization process of workstation screen image sequence, the palette's design method is optimized to enhance color quantitative speed. Experimental results show that the image reconstruction performance of proposed algorithm is better, and detail information of image is reserved. Furthermore, when it comes to continuous image sequences color quantization, the computing speed of proposed algorithm is three times faster than common color quantization method. As a complex image color quantization method, proposed method can not only adapt to workstation screen image, but also be used for other kinds of computer screen images.(4) Research on coding technology of workstation screen imageImage coding technology can effectively remove the redundant data of images, and is a important way to minimize image data quantity. Current image coding method does not consider the features of workstation screen image, and there are still large improvement space in image compression ratio of existing image coding algorithm. For such problems, this paper presents coding method of layered image compression. Firstly image layered operation is carried out as to transform an original image into a binary foreground image and colorful background image. And then the compression of foreground image is realized by the optimization of binary image algorithms. Thirdly, application of RPC encoding and RLE and LZO mixed coding algorithm is used for compression of background image. Experimental results show that, with similar coding speed as premise, the proposed algorithm increases compression ratio of the workstation screen image as 2-3 times as common image coding algorithms. As a complex image coding method, proposed method can not only adapt to workstation screen image, but also be used for other kinds of computer screen images.
|
PDF Full Text Request