In this paper, we derive a powerful algorithm for estimating missing line integrals in limited data computed tomography. We have formulated the estimation problem of missing line integrals in Radon space by the theory of projection onto convex sets (POCS) and utilized the Helgason-Ludwig consistency condition as a constraint in Radon space. Further, in order to stabilize the estimation process, we have used a priori knowledge about the rough shape of the object to be imaged. In the numerical experiments, it was confirmed that our method is superior to the previously reported algorithms from the view point of the computational cost and the image quality.
The complex autoregressive model we proposed as a shape descriptor has such a merit that the complex autoregressive coefficients are invariant for translation, rotation and dilatation of shapes. Those are also invariant for a start point of boundary tracking. In this paper, several distances for retrieving similar shapes are proposed based on the model. The distances can be seen to be natural extension of the distances based on the conventional autoregressive model. The properties of the distances are discussed, and effectiveness of the distances for shape retrieval is verified by some experiments.
Falling attitudes of snowflakes were smultaneously photographed by two video cameras from horizontal and vertical directions, and these images were analyzed by an image processor and personal computer. The attitude of each snowflake was measured every 1/30 second frog! its two images. It is generally found that snowflake images from the top are larger than the ones from the side and some of snowflakes rotate during fall.
In this paper, we describe image processing system with multiprocessor manner and MLIP (Multiprocessor Language for Image Processing). Out image processing system is constructed from one master processor and four processor units with DSP (Digital Signal Processor). MLIP is the system software which is executed on the host computer. Using this MLIP, we are efficiently able to write image processing program and control this image processing system. In comparison with typical personal computer, speedup ratio has become from three times to seventy-three times.
It has been known that the human visual system uses non-uniformly distributed cone cells. This paper presents a methode of understandig complex figures with straight lines by simulated the hu-man visual system. A equipment is consited a two-axis rotatable TV camera, a computer and a TV monitor. Video signals are changed into non-uniformly resolution signals like a eye. If the TV camera catches a watchedpoint of figure, then the TV camera is rotated to analyze the figure at the high resolution area. Characters of figures can be extracted and sored in files to understand some figure.