Object-oriented and primitives-based understanding of image processing in image processing

Object-oriented is the idea of modern software development, is the focus of the current IT industry, is the mainstream of software development methods since the 1990s. Object-oriented concepts and applications are not limited to programming and software development, and have been extended to a wide range.

Object-oriented image analysis technology is applied to the interpretation of remote sensing images even in the the 1970s. Ketting and Landgrebe (1976) presents the advantages of homogeneous object extraction, and proposes a segmentation algorithm called echo (extraction and classification of homogenous Objects). The object-oriented method is to divide the image into objects with different sizes, and each object has such attributes as spectrum, shape, texture, spatial topological relation and hierarchy. The object-oriented remote sensing image analysis method divides the remote sensing image into different homogeneity objects, each object has the attribute of describing the information such as spectrum, shape, structure, texture and so on, and there are some relationships between the objects, such as neighboring, including, and the inheritance of similar object-oriented software development. The analytic entity is no longer a single pixel, but a meaningful image object composed of several pixels, including the spectral, semantic and texture information of the image object. Object-oriented image analysis adopts a method of image multistage segmentation, which generates attribute information similar to Image polygon objects (Faber and forstner,1999) with arbitrary scale, uses fuzzy mathematics to obtain the attribute information of each image object, and takes the image object as the basic unit of information extraction, The purpose of realizing the automatic extraction of category information.

Reference documents:

[1] Mingdongping. Research on extraction and pattern discrimination of high resolution remote sensing feature primitives. 2006 Ph. D. thesis, Institute of Geography and Resources, CAS
[2] Huanghuiping. Research on scaling problem in object-oriented image analysis. 2003 Doctoral dissertation, Institute of Remote sensing applications, CAS