scispace - formally typeset
Search or ask a question

Showing papers on "Corner detection published in 1989"


Journal ArticleDOI
TL;DR: The effectiveness of the new method for detecting corners based on local symmetry of the shape is illustrated with the results of experiments for a number of different types of digital curves.

20 citations


Proceedings ArticleDOI
26 Mar 1989
TL;DR: The authors present a systematic procedure for testing corner detection algorithms used in digital image processing, and a previously proposed corner detection algorithm was tested using the proposed method.
Abstract: The authors present a systematic procedure for testing corner detection algorithms used in digital image processing. The characteristics of a corner are described by six different attributes: corner angle, corner arm length, corner adjacency, corner sharpness, gray-level distribution, and noise level. Elementary test images are used to test each attribute individually. A previously proposed corner detection algorithm was tested using the proposed method, and the results are presented. >

10 citations


Journal ArticleDOI
TL;DR: A computer vision based method to convert a manual font library on paper into concise computer representations through an automatic process to generate high quality Chinese fonts for desktop publishing through anautomatic process.

2 citations


Proceedings ArticleDOI
01 Jan 1989
TL;DR: This paper describes a technique to align or register tWO overlapping images using only the contents of the images using a set of minuria features computed for each image where the minutiae are defined as unique changes in direction in constant image intensity lines or small closed segmented regions.
Abstract: This paper describes a technique to align or register tWO overlapping images using only the contents of the images. A set of minuria features are computed for each image where the minutiae are defined as unique changes in direction in constant image intensity lines or small closed segmented regions. The resulting minutia patterns are then matched or correlated with the location of the correlation peak being the necessary translation needed to align the two images. The images must be approximately to the same scale. Rotations can be handled by successive application of the translation algorithm using specified rotations for one of the images. Motion of an object is detected from a simple comparison of its registered cwrdinates in the two images. The technique has been applied to many infrared and video images and works well where there is sufficient scene richness.

1 citations