Microarray image processing by order morphology

Abstract

An order morphology transform is presented to filter and segment which is done by different percentile. Filter is done flexibly by different size structure element with several percent. The threshold which for normal segment way such as Ostu decides is more lower when a low SNR image is processing and especially the foreground is small or dot. The foreground can not be identified in those case. Adaptive multi-threshold segment is defined by percent value of order morphology. Analysis and results indicate that this way is more adaptive to different SNR fluorescence images. It could be applied to process high-density chips.

References

[1] RadhakrishnanN, CharlotteA. Peterson. IEEE Transaction on NanoBioscience, 2002, 1: 78-78

[2] Roberto Hirata and Jr. Junior Barrera. Proceedings of the 14th Brazilian Symposium on Computer Graphics and image Processing (2001), 112.

[3] Laura R and SPIE MBER,1679 (1998), 16.

[4] CinqueL, LevialdiS, RosenfeildA. Pattern Recognition, 1997, 28: 901-901

[5] RamacL C, VarshneyP K. Pattern Recognition, 1998, 30: 1161-1161

[6] HojjatolestarniS A, KittlerJ. IEEE Transaction on Image Processing, 1998, 7: 1079-1079

[7] OtsuN. IEEE Trans Man Cybernet. SMC, 1978, 8: 62-62

[8] AndreB, JoshuaL L. Computer Vision and Image Understanding, 2000, 77(3): 317-370

[9] http://support.axon.com/