Adaptive edge detection and image filtering using light sensor

Edge detection plays a vital role in image processing. Detected edge can be used for feature analysis, pattern analysis, frequency analysis, and etc. however the result of this analysis subjects to changes in surrounding light intensity. In condition where surrounding light intensity behave dynamically, knowledge of its value is very important, so that the result of edge detection matches the actual light intensity. Thus, two techniques are proposed for this approach, first by using camera itself, and the second one by using a photo sensor. An edge detection algorithm should be able to be use for all images captured in various environment conditions. However, this is not the case because most edge operator values are fixed. For non-real-time application condition it is acceptable, but for real-time application in dynamic environment, it is not so. Such different in processing time performance will be investigated. Instead of generating the kernel values based on training images, our methods emphasize on generalized kernel weights according to actual light intensity.

PDF (Full text) 38997.pdf - Published Version Restricted to Repository staff only Download (97kB)

Actions (login required)

View Item