Abstract:
In the recent time, the problem in identifying criminals has been alarming and also
detrimental to the nation’s security. There is need for security enforcement agencies to
develop a strategy to ensure that “round pegs are put inside round hole” - innocent citizen
would not be crucified while the real criminals go free. In this research, image background
matching system was developed to ensure that justice prevails, so that real criminals are
identified accurately and brought to book. To implement matching model, image pixel
analysis of feature/key point was used to achieve a better quality result.
There is a need for additional work to enhance image background matching with reduced
keypoints/features to analyze pixels towards the improvement of matching accuracy using
SIFT algorithms. This research thereby provides a system which was able to work with
reduced keypoint/feature (Patch). Image Euclidean Distance (IMED) was adopted as a metric to guarantee the universal validity of the distance measure among pixels. The metric coefficients depend properly on the pixel distances obtained from features. The research makes use of image size or weight, image resolution as the dimension (image width and height) and image complexion or colour during the comparison process. Also, the percentage difference was obtained as the degree of non-resemblance between the images to be matched.
Therefore, for an image to be matched with the database the percentage difference must be below or exactly 20%, otherwise, any result above this range indicates mismatched or not matched. The research established an effective and efficient system towards identifying crime suspect thereby, reducing the rate of crime in society.
