Image analysis and image processing of shoeprints are essential ingredients of numerous tasks mostly related to recognition and identification, but also to image enhancement, data visualization and other objectives. A brief overview of these tasks with some examples will be given. The major part of the talk will focus on challenges in image analysis of shoeprints, we faced in developing the SESA (Statistical Evaluating of Shoeprints Accidentals) system. The lessons we learned will be presented as well as our understanding of the challenges ahead. Our main objective in the SESA project was to describe the probability of occurrence of accidental characteristics in shoe prints. Accidentals are (usually) small defects such as tears, cuts and nicks, caused to the shoe sole during the usage of the shoe. Accidentals present on the suspect’s shoe as they appear on the test impression and identified on crime scene prints as well, may provide enough information to establish identification of the shoe. In developing the system we faced both theoretical and practical challenges. For example, we had to decide how to define accidentals and how to mark them digitally. The challenges here include issues of segmentation, shape description and analysis, and texture modeling. In order to estimate the probability of occurrence of accidentals we marked numerous accidentals on multiple shoeprints, and measure three features of each accidental: its location, orientation and shape. This raised further questions of defining and measuring each feature: what is location of a characteristic and with respect to what should it be measured, how orientation should be defined and measured, and what is shape and how to measure it. Each question lead to more open tracks, as will be described. Combining the information from many shoes of various patterns raised the problems of alignment and normalization. Challenges here are defining and constructing a universal shoe aligned coordinate systems, and dealing with partial shoeprints. Noise is a major issue when building pattern recognition systems. The challenges here are measuring the noise of the different components of the system, including those that involved human operators. Issues of human-machine interaction and user interfaces that are tightly related to the image analysis system add complexity. In developing SESA we dealt mainly with controlled test impressions, but we gained insights on the broader problem, and that assisted in identifying the major challenges ahead: Dealing with variable inputs. The inputs to analysis systems may vary considerably, depending on the retrieval and acquisition methods, and whether the shoeprints are crime scene prints or lab impressions. Treating noise. Related to the large variations in inputs are the numerous sources of noise in shoeprints. Lack of information. In practice, shoeprints may lack information needed by the expert examiner to identify details without the shoe itself or at least images of the shoe sole. Can automated systems be developed to use only shoeprints? How should shoeprints and shoe sole images be used together? Scalability of recognition systems and databases of shoeprint accidentals. There are a lot of challenges in managing large systems that are supposed to be used by a community of researchers and practitioners. Some are related to image analysis, e.g. establishing standard tools of marking and analysis, and developing fast and accurate retrieval systems.