Occlusion Images Generation from Occlusion-Free Images for Criminals Identification based on Artificial Intelligence Using Image


  • Yonggeol LEE
  • Jang Mook KANG






Image Synthesis, Criminals Identification, Pixel Weights, Statistic


In this paper, we propose the method to generate occlusion images which include sunglass and scarf from occlusion-free images for criminals identification. First, we construct an auxiliary set that has occlusion-free face image and its occluded face images from AR database. Secondly, pixel weights are determined by the statistical analysis of the degree depending on whether occlusion-free or occlusion. As the weight is closer to 1, the pixel value of original image is maintained as much as possible. Therefore, pixel values of occlusion-free region are preserved when the weights are closed to 1. Third, pixel values of the occlusion region of original image are replaced to pixel values which are the mean of occlusion images by multiplying the composited weights. As a result of generate face images with various occlusions using the proposed method, the images were able to preserve the person-specific characteristic, and criminals identification has become easier.




[1] Arai, M., Okumura, K., Satake, M. and Shimizu, T., Proteome-wide functional classification and identification of prokaryotic transmembrane proteins by transmembrane topology similarity comparison. Protein Sci., 13 (2004), 2170-2183.

[2] Saier, M. H., Jr., A Functional-Phylogenetic Classification System for Transmembrane Solute Transporters. Microbiol. Mol. Biol. Rev., 64 (2000), 354-411.

[3] Shafrir, Y. and Guy, R., STAM: simple Transmembrane Alignment Method. Bioinformatics, 20 (2004), 758-769.

[4] Rost, B., Twilight zone of protein sequence alignments. Protein Eng., 12 (1999), 85-94.

[5] Yang, A.-S. and Honig, B., An Integrated Approach to the Analysis and Modeling of Protein Sequences and Structures. II. On the Relationship between Sequence and Structural Similarity for Proteins that are Not Obviously Related in Sequence. J. Mol. Biol., 301 (2000), 679-689.

[6] Boeckmann, B., Bairoch, A., Apweiler, R., Blatter, M.-C., Estreicher, A., Gasteiger, E., Martin, M. J., Michoud, K., O'Donovan, C., Phan, I., Pilbout, S., and Schneider, M., The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003. Nucleic Acids Res., 31 (2003), 365-370.

[7] Sugiyama, Y., Polulyakh, N. and Shimizu, T., Identification of transmembrane protein functions by binary topology patterns. Protein Eng., 16 (2003), 479-488.

[8] Lao, D. M. and Shimizu, T., A Method for Discriminating a Signal Peptide and a Putative 1st TM Segment. 2001 International Conference on Mathematics and Engineering Techniques in Medicine and Biological Sciences (METMBS '01). CSREA Press, Las Vegas, 2001, pp. 119-125.

[9] Arai, M., Mitsuke, H., Ikeda, M., Xia, J.-X., Kikuchi, T., Satake, M., and Shimizu, T., ConPred II: a consensus prediction method for obtained transmembrane topology models with high reliability. Nucleic Acids Res., 32 (2004), W390-W393.

[10] Thompson, J. D., Higgins, D. G. and Gibson, T. J., CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res., 22 (1994), 4673-4680.

[11] Arai, M., Fukushi, T., Satake, M. and Shimizu, T., A proteome-wide analysis of domain architectures of prokaryotic single-spanning transmembrane proteins. Comput. Biol. Chem., 29 (2005), 379-387.

[12] Sonnhammer, E. L. L., Eddy, S. R. and Durbin, R., Pfam: A Comprehensive Database of Protein Domain Families Based on Seed Alignments. Proteins, 28 (1997), 405-420.

[13] Liu, Y., Gerstein, M. and Engelman, D. M., Transmembrane protein domains rarely use covalent domain recombination as an evolutionary mechanism. Proc. Natl. Acad. Sci. USA, 101 (2004), 3495-3497.

View Full Article: