An image processing architecture for real time generation of scale and rotation invariant patterns

Messner, Richard A.; Szu, Harold

doi:10.1016/s0734-189x(85)80075-x

Cited by 63 publications

(12 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This issue has been recently revived in a modified guise through the proposal of a neural network model of visual pattern rotation (Goebel, 1990; see also Morgan 1983). In the model it is proposed that the orientation invariance of complex stimuli is achieved through a piecemeal rotation of the line components and not through a holistic rotation of their analogue images as suggested by other formal, albeit physiologically unrealistic, models (Messner and Szu 1985;Reitboek and Altman 1984;Sheng 1989). An actual demonstration of a rotation effect with simple line stimuli would clearly add some psychological credence to the network proposal that already has received some, though indirect, neurophysiological support (Georgopoulos et al 1989;Pellizer and Georgopoulos 1993).…”

Section: Imentioning

confidence: 99%

Mental-Rotation Effect: A Function of Elementary Stimulus Discriminability?

et al. 1996

View full text Add to dashboard Cite

Abstract. It is well known that when humans have to decide whether two differently oriented shapes are identical or mirror images their performance deteriorates as a function of the orientation disparity (mental-rotation effect). Here it is shown that the effect can also be obtained reliably with non-mirror-image, arbitrarily different polygons provided they are previously selected to be hard to discriminate. The slope of the decision speed versus orientation disparity functions was found to be inversely related to the discriminability of shapes under conditions of no, ie 0°, orientation disparity. Easily discriminable polygon pairs yielded essentially flat, no-effect functions. The arbitrary polygons that were more difficult to discriminate produced a rotation effect that was similar to those of mirror-image polygon pairs. Mirror images in this context may only be a special case of hard-to-discriminate stimuli. We also show that the speed of judging whether simple lines were of the same or different length was similarly subject to a rotation effect provided that the length differences were sufficiently small, ie when their baseline dicriminability was poor enough. It is suggested that the mental rotation of complex shapes (eg polygons) may build on rotation effects pertaining to the simpler elements of which they are composed. Further, some special eflects associated with the rotation of such simpler components may explain certain peculiarities apparent in orientation invariance functions obtained with complex stimuli.

show abstract

Section: Imentioning

confidence: 99%

Mental-Rotation Effect: A Function of Elementary Stimulus Discriminability?

et al. 1996

View full text Add to dashboard Cite

show abstract

“…(20,26) The program failed to recognize a 90 degree rotation and 50% scaling (see Table VI and 0.0 14.0 (9, 10) to (9,24) 85.0 9.0 (9, 24) to (18,24) 85.0 7.0 (10,18) to (17,18) for the normal and 5 degrees for the angle (see Table VIIl). (15,12) to (15,18) 70.0 27.5 (4,14) to (30,5) 70.0 24.4 (0, 15) to (23,7) 105.0 9.5 (21,22) to (30,25) 110.0 30.7…”

Section: Simulation Resultsmentioning

confidence: 99%

“…Many methods have been developed for this purpose, such as moment invariant method [13], dynamic alignment [14] and complex log mapping related methods [15,16]. So far I have not seen invariant recognition algorithms using the Hough Transform and its generalization.…”

Section: Introductionmentioning

confidence: 99%

Invariant pattern recognition algorithm using the Hough Transform

Li¹

2000

View full text Add to dashboard Cite

A new algorithm is proposed which uses the Hough Transform to recognize two dimensional objects independent of their orientations, sizes and locations. The binary image of an object is represented by a set of straight lines. Features of the straight lines, namely the lengths and the angles of their normals, their lengths and the end point 2 positions are extracted using the Hough Transform. A data structure for the extracted lines is constructed so that it is efficient to match the features of the lines of one object to those of another object, and determine if one object is a rotated and/or scaled version of the other. Finally a generalized Hough Transform is used to match the end points of the two sets of lines. The simulation experiments show good results for objects with significant linear features .

show abstract

“…For example, the complex log transformation which approximates the mammalian retino-cortical mapping is also widely used as a means for constructing space-variant vision systems 1 (Rojer and Schwartz, 1990;Weiman, 1988;Dario, 1989, Sandini et al, 1989;Messner and Szu, 1986;Schenker et al, 1981;Bonmassar and Schwartz, 1994;Bonmassar and Schwartz, 1995;Yamamoto et al, 1996). In this approach, pixel sampling is performed at exponentially increasing intervals.…”

Section: Originalmentioning

confidence: 99%