The complete analysis and interpretation of the information in image data is a complex process. This dissertation presents 3 major contributions to image analysis, namely, global multiscale detection, local scale analysis, and boundary extraction. Global scale analysis is related to identification of the various scales presented in the image. A new approach for global scale analysis is developed based on the differential power spectrum normalized variance ratio (DPSNVR). The DPSNVR is the ratio of the second order normalized central moment of the power spectrum of the image to that of the multiscale differential mask. Local maxima in DPSNVR graph directly indicate the global scales in the image. Local scale analysis performs a more detailed analysis of the edges to eliminate effects of blurring. A method based on mutilscale feature matching has been proposed. Details obtained at all scales are treated using a scale invariant normalization scheme. Besides local scale analysis, a multiscale data fusion algorithm has been implemented which leads to the new concept of multiple scale differential masks. The multiple scale differential mask generated using a range of scale values possesses the remarkable shape preservation property which makes it superior to traditional multiscale masks. Finally the complete sequential boundary extraction algorithm based on particle motion in a velocity field is presented. The boundary extraction algorithm incorporates edge localization, boundary representation, and automated selection of boundary extraction parameters. The global scale analysis techniques in conjunction with the boundary extraction algorithm provide a multiscale image segmentation algorithm.