VideoComparison (version 0.15)

VideoMatch: Starting from two list of image hashes and the score in motion matching factor.

Description

VideoMatch looks for the likelihood of coherence between the sequence of pertinent image hashes obtained after preparing the comparison between video Motion vector.

Likelihood estimation relates the coherence (distance: norlized average of four normalized distances) between image sequence from the two videos in the region where their motion curve matches. It can be affected by the factor of coherence of this matching.

Usage

VideoMatch(lh1,lh2,sc)

Arguments

lh1

List with the ordered sequence on hashes from images belonging to the common motion chunck. It refers to the video1.

lh2

List with the ordered sequence on hashes from images belonging to the common motion chunck. It refers to the video2.

sc

This is the motion scoring factor. It can be derived from the VideoComparison function (first argument of the answer).

References

Espinosa-Gutiez, S., Ordieres-Mere, J., Bello-Garcia, A.: Large scale part-to-part video matching by a likelihood function using featured based video representation. TAMIDA 2013 - Taller de Mineria de Datos dentro del IV congreso Espannol de Informatica. 254-257 (2013). http://bioinspired.dacya.ucm.es/maeb2013/images/ActasCAEPIA_final.pdf

Cui, M., Femiani, J., Hu, J., Wonka, P., Razdan, A.: Curve matching for open 2D curves. Pattern Recogn. Lett. 30, 1-10 (2009)

Examples

#Load json data from txt
fileName1 = strcat(path.package(package="VideoComparison"),"/extdata/imagehash1.gz")
json1 = readLines(gzfile(fileName1))
fileName2 = strcat(path.package(package="VideoComparison"),"/extdata/imagehash2.gz")
json2 = readLines(gzfile(fileName2))

# Extract data from json
out1<-fromJSON(json1)$hits$hits[[1]]$`_source`$Hash;
dct1<-as.character(out1[1])
hstrada1<-as.numeric(unlist(strsplit(out1[2],",")))
mw1<-as.numeric(unlist(strsplit(out1[3],",")))
rd1<-as.numeric(unlist(strsplit(out1[4],",")))
img1<-list(dct=dct1,hstrada=hstrada1,mw=mw1,rd=rd1)

out2<-fromJSON(json2)$hits$hits[[1]]$`_source`$Hash;
dct2<-as.character(out2[1])
hstrada2 <-as.numeric(unlist(strsplit(out2[2],",")))
mw2<-as.numeric(unlist(strsplit(out2[3],",")))
rd2<-as.numeric(unlist(strsplit(out2[4],",")))
img2<-list(dct=dct2,hstrada=hstrada2,mw=mw2,rd=rd2)

## Requires specific server architecture
# img1<-ExtractImgHash("000060","C0031D0")
# img2<-ExtractImgHash("000009","C0035D0")

VideoMatch(img1,img2)