[Pattern recognition]. (Greece) ritis <version 4> Note 8 _ template matching

For speech recognition, the same word is said differently by the same person each time, which brings difficulties to recognition. This chapter discusses how to define metrics that adapt to different features in different situations.

1. Measure Based on Optimal Path search: ① bellman's optimum principle and Dynamic Programming ② the edit distance ③ In the speech recognition Dynamic Time distortion (DTW), speaker-dependentrecognition. speaker-independentrecognition.

2. Correlation-based measurement: This part solves the problem of "giving a set of record data, finding whether the data contains a known pattern, and finding its specific location ". Many applications involve this issue: Target Detection, robot vision, and video encoding.

3. extensible template model: This part aims to seek template matching to explain the differences between the reference template and the test template in the image.

4. Content-based Information Retrieval: relevance feedback. Traditional Information Retrieval is text-based. This section describes Content-based information retrieval, for example, image retrieval, by Using Image Textures, colors, and shapes.

[Pattern recognition]. (Greece) ritis <version 4> Note 8 _ template matching