Enhancement of seminal stains using background correction algorithm with colour filters

https://doi.org/10.1016/j.forsciint.2016.03.046Get rights and content

Highlights

  • No image processing has been applied on seminal stains in literature.

  • A new background correction algorithm (BCA) for seminal stains is proposed.

  • Processing on two seminal stains images: with and without colour filters.

  • Significant improvement on the detectable dilution after applying our BCA.

  • This approach is simpler and cheaper than utilizing band-pass filters for BCA.

Abstract

Evidence in crime scenes available in the form of biological stains which cannot be visualized during naked eye examination can be detected by imaging their fluorescence using a combination of excitation lights and suitable filters. These combinations selectively allow the passage of fluorescence light emitted from the targeted stains. However, interference from the fluorescence generated by many of the surface materials bearing the stains often renders it difficult to visualize the stains during forensic photography. This report describes the use of background correction algorithm (BCA) to enhance the visibility of seminal stain, a biological evidence that fluoresces. While earlier reports described the use of narrow band-pass filters for other fluorescing evidences, here, we utilize BCA to enhance images captured using commonly available colour filters, yellow, orange and red. Mean-based contrast adjustment was incorporated into BCA to adjust the background brightness for achieving similarity of images’ background appearance, a crucial step for ensuring success while implementing BCA. Experiment results demonstrated the effectiveness of our proposed colour filters’ approach using the improved BCA in enhancing the visibility of seminal stains in varying dilutions on selected surfaces.

Introduction

Appropriate visualization and detection of colourless biological stains such as seminal, saliva and urine at crime scene level is crucial for initiating further laboratory analysis to solve the crime [1]. Fluorescence imaging is a technique that enables visualization of untreated biological stains like semen [2], [3], saliva and urine [3] as well as chemically treated trace evidences such as fingerprint and bloodstains [4]. Fluorescence is characterized by the absorption of light of shorter wavelength (excitation spectrum) by a substance and emission of light of a longer light wavelength (emission spectrum) [5].

Nevertheless, fluorescence (emission light) of seminal stains is considerably weaker compared to the excitation light. Therefore, light filters are often used to block the excitation light and permit only the seminal stains’ emitted light during crime scene investigation [5]. For instance, fluorescence from seminal stains can be viewed through yellow or orange filters when illuminated by blue light. However, varying degree of fluorescence from the background materials can interfere with the fluorescence from the seminal stains rendering the latter less distinguishable. In this matter, a procedure for scanning the stained area using light of multiple wavelengths and filters was introduced [2]. Such approaches mainly used yellow, orange and red filters, which usually part of the forensic light source (FSL) and routinely available in crime scene investigation kits.

The advantages of using background reduction imaging techniques such as background correction algorithm (BCA) for enhancing visualization of diluted bloodstains has been reported [6], [7]. BCA is effective in enhancing the visibility of diluted bloodstains, which have a strong narrow peak around 415 nm [2]. The effect of BCA in enhancing fluorescence from chemically treated fingermarks and bloodstains has also been studied [4]. This algorithm uses the spectral information from the stains when illuminated by light of two or three specific light wavelengths or when specified band-pass filters are used during photography. Wagner and Miskelly [6], [7] utilized illuminations of varying light wavelengths such as 395 nm, 415 nm and 435 nm to perform BCA and reported that visibility of bloodstains was further improved after processing. The use of BCA can also improve the visibility of bloodstains using images illuminated with 415 nm and blue light (LED-based illuminations) that are commonly available in forensic light sources (FLS) [8].

For transmittance/absorption mode, BCA [6], [7] is calculated asIout=2I1I2+I3×200128where I1 is the intensity of peak absorption and I2 and I3 are the light wavelengths of the two neighbouring images respectively. For instance, I1, I2 and I3 are images of 415 nm (peak absorption), 395 nm and 435 nm (another two neighbouring wavelengths) respectively, as reported by Wagner and Miskelly [6], [7]. Note that I3 is substituted as I2 for two-wavelength BCA, which resulted the calculation of the division part as I1/I2.

For fluorescence mode, BCA [4] is calculated asIout=I1I2+I32where I1, is the intensity of peak emission and I2 and I3 are the light wavelengths of the two neighbouring images, respectively. Note that I3 is substituted as I2 for two-wavelength BCA which resulting the calculation as I1  I2. The output of this equation is adjusted to 0–255 manually for best viewing range.

The equations above do not include adjustment of brightness or exposure between images. In fact, this algorithm would fail if the brightness of image background differs. Therefore, it is crucial to ensure similarity in the brightness of the background among different images so that background in the resultant images would be as nearer to 1 and 0 as possible for transmittance and fluorescence, respectively. Note that best resultant images for transmittance mode would be stains appear dark and backgrounds appear bright, while it is the opposite for fluorescence mode where stains appear bright and backgrounds appear dark.

In this paper, we focus on the fluorescing property of seminal stains and the enhancement of seminal stains’ visualization is performed through BCA in fluorescence mode (Equation (2)). Earlier reports have shown the successful use of BCA in improving the visibility of other fluorescing stains (none are about seminal stains) when using two or three narrow light wavelength band-pass filters. Furthermore, BCA has not been tested to discriminate fluorescing stain images captured using the commonly available colour filters in FLS kit.

Therefore, instead of using specific band-pass filters of narrow light wavelength, we implemented BCA on images captured with commonly used colour filters in forensic applications, namely yellow, orange and red. Two-wavelength BCA was utilized on two images of seminal stains: one captured with one of the aforementioned filters and the other captured without any filter. In fact, auto-exposure setting of camera could not ensure similar exposure or brightness for the images captured under the above two conditions. Therefore, we incorporated mean-based contrast adjustment (mCA) as pre-adjustment step on images prior to BCA, which was proven to be successful for enhancing different brightness images of bloodstains through absorption mode in our previous reported work [9]:Iout=IinImean×Itargetwhere Iout is the pixel intensity value of the output image, Iin is the pixel intensity value of the input image, Imean is the mean intensity of the image and Itarget is the brightness value of the targeted output that will be set as 1 in our algorithm.

The mCA is used to adjust the background brightness of images to be similar through division of their corresponding mean values. Therefore, our improved version of BCA, namely mean-based adaptive background correction algorithm (mABCA), is considered as more robust in processing images with background that differ in brightness. Compared to our previous two works for BCA [8], [9], where both are reported for bloodstains through BCA in absorption mode (Equation (1)), we proposed novel approach for BCA using colour filters for seminal stains through mABCA in fluorescence mode in this study.

Section snippets

Theory

Fluorescent emission spectra of seminal stains have been shown to have light wavelength peak at around 450 nm and 510 nm under the excitation light wavelength ranges between 350 nm and 450 nm, respectively [2]. However, only three emission wavelengths were mentioned earlier [2]. In this research, we tested the spectral fluorescent characteristics of seminal stains using excitation light wavelength of wider range from 330 nm to 600 nm. Perkin Elmer LS55 Spectrofluorometer (PerkinElmer Inc.) was used

Seminal stains preparation

Seminal samples used in this study were obtained from Hospital Universiti Sains Malaysia (USM) upon ethical approval by Human Research Ethics Committee of Universiti Sains Malaysia (USM). The seminal samples were serially diluted to yield dilutions of 1×, 2×, 5×, 10×, 25×, 50×, 100× and 250×. A small single drop of each samples was stained onto five porous surfaces of different colours: white cotton, light blue cotton, green cotton, blue denim and light brown wood; as well as five non-porous

mABCA application in improving seminal stains visibility

Table 1, Table 2 show the summary of the results pertaining to detection of seminal stains on the porous surfaces and non-porous surfaces, respectively. Most of the non-porous surfaces indicated high detection rate compared to porous surfaces, where higher dilution can be detected. This can be explained as due to the absorption of portion of the seminal stain into the surfaces of the porous materials which does not happen for non-porous surfaces. Yellow filter was effective only for the

Conclusions

This research report introduces mABCA as new technique for enhancing the visibility of fluorescing seminal stains in images captured with commonly available colour filters in forensic field: yellow, orange and red. The results demonstrate the effectiveness of mABCA on colour filters in enhancing the visibility of more diluted seminal stains which are difficult to be visualized even with the aid of colour filters. Notably, the method described here permits the use of colour goggles by placing

Acknowledgements

Great appreciation to Assoc Prof P.T Jayaprakash for his insightful comments to improve this manuscript. The authors would also like to thank Hospital Universiti Sains Malaysia for providing the seminal sample in completion of the research work. This work is supported by Universiti Sains Malaysia through Postgraduate Fellowship Scheme, RU grant [1001/PELECT/814204] and Ministry of Higher Education, Malaysia through PRGS grant [203/PELECT/6740017].

References (14)

There are more references available in the full text version of this article.

Cited by (3)

View full text