Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

Documentation on the match shapes return result

I have been looking through the OpenCV documentation and I cant find any mention of the cv::matchShapes return value. There seem to be three methods cv::CONTOURS_MATCH_I1, cv::CONTOURS_MATCH_I2, cv::CONTOURS_MATCH_I3. They are a sum of distances so I guess a perfect match is 0 however what is the possible value range (maybe its not possible to give this - but cv::CONTOURS_MATCH_I3 seems to normalise?). In either case the documentation should give information on at least the fact that a perfect match has a result of zero.

Documentation on the match shapes return result

I have been looking through the OpenCV documentation and I cant find any mention of the cv::matchShapes return value. There seem to be three methods cv::CONTOURS_MATCH_I1, cv::CONTOURS_MATCH_I2, cv::CONTOURS_MATCH_I3. They are a sum of distances so I guess a perfect match is 0 however what is the possible value range (maybe its not possible to give this - this, but cv::CONTOURS_MATCH_I3 seems to normalise?). scale the moments at least?). In either case the documentation should give information on at least the fact that a perfect match has a result of zero.

The thing is if you want to look for matches better than a certain score, which score can you choose? IMHO the scores should be normalised somehow eg. 0-100 or 0-1.

Documentation on the match shapes return result

I have been looking through the OpenCV documentation and I cant find any mention of the cv::matchShapes return value. There seem to be three methods cv::CONTOURS_MATCH_I1, cv::CONTOURS_MATCH_I2, cv::CONTOURS_MATCH_I3. They are a sum of distances so I guess a perfect match is 0 however what is the possible value range (maybe its not possible to give this, but cv::CONTOURS_MATCH_I3 seems to scale the moments at least?). In either case the documentation should give information on at least the fact that a perfect match has a result of zero.

The thing is is, if you want to look for matches better than a certain score, which score can you choose? IMHO the scores should be normalised somehow eg. 0-100 or 0-1.

Documentation on the match shapes return result

I have been looking through the OpenCV documentation and I cant find any mention of the cv::matchShapes return value. There seem to be three methods cv::CONTOURS_MATCH_I1, cv::CONTOURS_MATCH_I2, cv::CONTOURS_MATCH_I3. They are a sum of distances so I guess a perfect match is 0 0, however what is the possible value range (maybe range? maybe its not possible to give this, but cv::CONTOURS_MATCH_I3 seems to scale the moments at least?). least. In either case the documentation should give information on at least the fact that a perfect match has a result of zero.

The thing is, if you want to look for matches better than a certain score, which score can you choose? IMHO the scores should be normalised somehow eg. 0-100 or 0-1.

Documentation on the match shapes return result

I have been looking through the OpenCV documentation and I cant find any mention of the cv::matchShapes return value. There seem to be three methods cv::CONTOURS_MATCH_I1, cv::CONTOURS_MATCH_I2, cv::CONTOURS_MATCH_I3. They are a sum of distances so I guess a perfect match is 0, however what is the possible value range? maybe its not possible to give this, but cv::CONTOURS_MATCH_I3 seems to scale the moments at least. In either case the documentation should give information on at least the fact that a perfect match has a result of zero.

The thing is, if you want to look for matches better than a certain score, which score can you choose? IMHO the scores should be normalised somehow eg. 0-100 or 0-1.0-1.

(It seems that log moments are better for matching purposes)