計算機視覺

Computer Vision   

　　Computer vision, like image processing, deals with digital images, but it differs greatly in its goals in the techniques that it uses．Image processing is primarily concerned with image-to-image operations，both the input and the output are images. In computer vision, on the other hand, the input still consists of images, but the goal is now to construct adscription of the scene from which the images were obtained. Such a description usually involves recognizing objects that are present in the scene and de-termini their properties and relationships The tech-inquest used to achieve this goal come primarily from from pattern recognition and artificial intelligence, rather than from signal processing．   

　　Computer vision is still, in some respects, more of an art than a science, Standard signal processing approaches are not sufficient to handle the range of problems and tasks encountered in vision. Advances in pattern recognition and artificial intelligence have led to most of the progress in the field, but major advances are still needed if significant further progress is to be achieved. The field of computer vision had its beginning over 25 years ago, and has let to many practical applications, but it still faces many challenges．For example, in the area of document processing, systems that read printed characters have been commercially available for many years，but reading unconstrained handwriting is still a research problem. In industrial applications，systems for various types of simple inspection task(e.g. Checking the alignment of into-grated circuit chip) are already in use, but systems that can allow a robot to pick parts out of a bin aren’t yet practical. Work is needed on the closer integration, and on the development of "expert systems that can make use of problem domain specific knowledge in controlling the application of these techniques to given visual tasks．

翻譯：

計算機視覺    

　　計算機視覺與圖象處理一樣 也是處理數字圖象，但兩者在所要達到的目的與所采用的技術方面差別很大。圖象處理主要關心圖象到圖象的運算，即輸入與輸出都是圖象。計算機視覺卻不一樣，輸入仍是由圖象組成，但其目的是形成一個對景物描述，輸入的圖象就是從此景物中獲得 的。這樣的一個描述通常涉及到對景物中物體的識別和確定它們的特性和相互關系。用來達到這種目的的技術主要來自模式識別和人工智能，而不是信號處理。在某些方面，計算機視覺仍是一種技巧，還不完全是一門科學標準的信息處理方法對視覺中所涉及的多種多樣的問題與任務來說是不能勝任的。而模式識別與人工智能的進展已導致了計算機視覺領域中的多項進步，但是，如果要獲得進一步有意義的進步，還需要模式識別和人工智能有一重大進展。
   
　　計算機視覺這個領域起始于25年前，至今己有多種實際應用，但是它仍面臨許多問題等待解決。例如，在文字資料處理方面，很多年前就可一從市場上溝得能閱讀印刷字符的系統，但是，閱讀小規則的手寫字符仍是一個研究課題。在工業應用中，完成各種類型的簡單檢查任務（例如，檢查集成電路芯片對準）的系統早已投入使用，但是使機器人從箱子里把零件取出來的系統尚未實用。今后還需要做工作，把信號處理技術與模式識別技術更緊密地結合起來以及開發“專家系統”，這種專家系統在控制這些技術應用于給定的視覺任務中能夠利用問題范疇的特定知識。

原文轉自：http://www.kjueaiud.com