Pengimejan salur darah tangan ialah satu pengkhususan dan pengetahuan yang amat
biasa dalam bidang biometrik dan perubatan. Ia didefinisikan sebagai suatu proses
perolehan imej, diikuti dengan manipulasi data imej untuk pelbagai tujuan dan aplikasi.
Kebelakangan ini, banyak kajian mencadangkan pelbagai cara untuk menghasilkan imej
salur darah tangan yang berkualiti. Namun, beberapa algoritma didapati agak mudah dan
gagal meningkatkan kualiti imej. Sebagai contoh, masalah pencahayaan yang tidak seragam
pada latar belakang imej akibat kelengkungan semula jadi tangan serta banding yang
rendah telah menimbulkan pelbagai isu. Oleh itu, projek ini bertujuan mencadangkan
teknik-teknik pemprosesan salur darah tangan yang berupaya menangani pelbagai masalah
tersebut tadi. Kaedah yang dicadangkan terutamanya terdiri daripada empat peringkat, iaitu
Penapisan Hingar, Peningkatan Tahap Skala Kelabu, Segmentasi Salur Darah dan
Peningkatan Binari. Pada peringkat pertama, imej tertakluk kepada penapisan ruang urutan
pangkat untuk penyingkiran hingar sebelum imej tersebut ditingkatkan dengan teknik
peningkatan gabungan AHEA dan CLAHE. Berikutan itu, morfologi Transformasi topiatas
digunakan untuk pembetulan pencahayaan latar belakang. Kemudian, Banding
Regangan meningkatkan banding salur darah ke tahap maksimum sebelum memasuki fasa
segmentasi Otsu ambang. Akhir sekali, ralat pada blok-blok piksel diperbetulkan
menggunakan Pembukaan and Penutupan Morfologi. Keputusan statistik, carta dan angka
menunjukkan bahawa kaedah yang dicadangkan telah membawa kepada peningkatan ketara
kualiti imej sementara mengekalkan PSNR dan peratusan entropi yang cukup tinggi.
_______________________________________________________________________________________________________
Hand vein imaging is common specialism and knowledge in the fields of vein-based
biometrics as well as medicine. It is defined as a process of image acquisition followed by
manipulation of the image data for various purposes and applications. In recent past, there
have been uncountable researches on vein imaging, proposing various image processing
approaches in obtaining good quality hand vein images. However, some algorithms are
found to be relatively simple in providing good enhancements of the vein image. For
instance, non-uniform background illumination problem due to natural curvature of human
hands as well as low contrast of veins have been some major. Hence, this project aims at
proposing a set of hand vein image processing techniques which is capable of handling
various aforementioned problems. The proposed method mainly comprises four stages,
namely Noise Filtering, Grayscale Enhancement, Vein Segmentation and Binary
Enhancement. The hand vein image captured is first subjected to Rank-order Spatial
Filtering for impulse noise removal before entering subsequent stage in which the image
contrast is improved by a combinational histogram-based enhancement technique,
constituted by AHEA and CLAHE. Following that, Morphological Top-hat Transformation
is employed for background illumination correction. Then, Contrast Stretching serves to
enhance the vein contrast to a maximum extent before entering the phase of Otsu
Thresholding segmentation. Last but not least, faulty pixels and blocks of the segmented
binary vein image are corrected using Morphological Opening and Closing. Statistical
findings, results, charts and figures show that the proposed method has led to significant
improvement of image quality while maintaining considerably high PSNR and entropy
percentage.