Visual Identification of People by Computer
Department of Computer Science, Stanford University., 1970 - Artificial intelligence - 476 pages
The thesis describes a computer program which performs a complex picture processing task. The task is to choose, from a collection of pictures of people taken by a TV camera, those pictures that depict the same person. The primary purpose of this research has been directed toward the development of new techniques for picture processing. (Author).
What people are saying - Write a review
We haven't found any reviews in the usual places.
A A A A A algorithm analysis applied approximate Artificial Intelligence attempt average background body CALL chapter character classification complete considered contains COORD coordinates dark area dark points dark region defined described detail detection determine developed difference direction discussed distance edge edge detection EEEE EEEEE examine example face Figure follows given gives goal head height horizontal identified included individual input intensity iris known left right light matrix measurements methods mouth necessary neighbor noise nose nostrils objects obtained operator original outline pattern performed person picture processing planning points position predicted present previously problem procedure reason recognition reduced region Report represent seconds shows sides smoothing square Stanford step structure success Table techniques template thesis threshold vector width ССС
Page 7 - A heuristic (heuristic rule, heuristic method) is a rule of thumb, strategy, trick, simplification, or any other kind of device which drastically limits search for solutions in large problem spaces. Heuristics do not guarantee optimal solutions; in fact, they do not guarantee any solution at all; all that can be said for a useful heuristic is that it offers solutions which are good enough most of the time.
Page 41 - ... objects they represent is needed. This is the idea behind the development of linguistic picture-processing techniques. A survey of these techniques can be found in Miller and Shaw (1968). Another approach to incorporating global structure into edge detection is the decision tree in use at SRI (Nilsson 1969). The global methods for edge detection also encounter problems. Here the program looks for specific edges in specific relationships. The search space is very large, tens of thousands of points...
Page iii - Health and in part by the Advanced Research Projects Agency of the Office of the Secretary of Defense (F44620-70-C-0107) which is monitored by the Air Force Office of Scientific Research.
Page 41 - ... scene would be relatively straightforward. However, the line drawing often contains flaws that seriously complicate its analysis. Some of these flaws could be corrected by more elaborate local processing. However, there is a limit to how well local processing can perform, and when significant edges cannot be told from insignificant edges on the basis of local criteria, the goal of producing a perfect line drawing in this way must be abandoned.
Page ii - I certify that I have read this thesis and that in my opinion it is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy.
Page 38 - ... some modest level of problem difficulty. Only schemes which actively pursue an analysis toward obtaining a set of sequential goals can be expected to extend smoothly into increasingly complex problem domains. Perhaps the most straightforward concept of planning is that of using a simplified model of the problem situation. Suppose that there is available, for a given problem, some other problem of "essentially the same character" but with less detail and complexity.
Page 39 - ... planning is that of using a simplified model of the problem situation. Suppose that there is available, for a given problem, some other problem of "essentially the same character" but with less detail and complexity. Then we could proceed first to solve the simpler problem. Suppose, also, that this is done using a second set of methods, which are also simpler, but in some correspondence with those for the original. The solution to the simpler problem can then be used as a "plan
Page 41 - The procedure for extracting edges must know what an "important" edge is and, therefore, it needs global information about the edges in the picture and the structure of the objects they represent.
Page 38 - ... centrality of the different candidates for attention. Following subproblem selection (for which several heuristic methods are proposed), one must choose methods appropriate to the selected problems. But for really difficult problems, even these step-by-step heuristics for reducing search will fail, and the machine must have resources for analyzing the problem structure in the large — in short, for "planning.
Page 73 - ... met, the head has been found. The sides of the head are then examined for indentations where the ears should be since the ears sometimes merge with the background. If indentations exist, these are filled in. Use plan to find outline in original This part of the program is a plan follower. Its input is the full size intensity picture and the list structure containing the small head outline. The output of the plan follower will be a new list structure containing the coordinates of an accurate outline...