Determine Conjugate Points of an Aerial Photograph Stereopairs using Seperate Channel Mean Value Technique

Abstract

In the development of digital photogrammetric system, automatic image matching process play an important role. The automatic image matching is used in finding the conjugate points of an aerial photograph stereopair automatically. This matching technique gives quite significant contribution especially in the development of 3D photogrammetry in an attempt to get the exact and precise topographic information during the stereo restitution. There are two image matching methods that have been so far developed, i.e. the area based system for gray level environment and the feature based system for natural feature environment. This research is trying to implement the area based matching with normalized cross correlation technique to get the correlation coefficient between the spectral value of the left image and its pair on the right. Based on the previous researches, the use of color image could increase the quality of matching. One of the color image matching technique is known as Separate Channel Mean Value. In order to be able to see the performance of the technique, a number of sampling areas with various different characteristics have been chosen, i.e. the heterogeneous, homogeneous, texture, shadow, and contrast. The result shows the highest similarity measure is obtained on heterogeneous sample area at size of all reference and search image, i.e. (11 pixels x 11 pixels) and (23 pixels x 23 pixels). In these area the correlation coefficient reached more than 0.7 and the highest percentage of similarity measure is obtained. The average of total similarity measure of conjugate images in the sampling image area only reach about 41.43 % of success. Therefore, this technique has a weakness and some treatment to overcome the problems is still needed.