An analysis of object-based intelligent image processing and retrieval system

In order to improve the process of analysis and retrieval of images, it is necessary to examine the execution of such program at the lowest level. This paper reports the results obtained from profiling the execution of an object-oriented image processing and analysis program termed ImageJ. Although profiling has been used in software engineering to identify execution bottlenecks, to our knowledge, it has not been considered as a means of analysis object based distributed image processing systems. The paper summaries the characteristics of the program in four aspects: classes and method codes profiling, method calls, parameter types and return types and CPU percentage. The profiling results show that the total amount of classes that was loaded at runtime is invariant to the change of the image size. Similar behaviours are observed for parameter types and return types. Based on the information, an intelligent monitor can be used to carry out automatic load scheduling and balancing